Semiotics in Information Systems Engineering

This page intentionally left blank
SEMIOTICS IN INFORMATION
SYSTEMS ENGINEERING
xrcnrNc ii i holds a chair of Computing Science at the School of Computing,
Staordshire University. He leads a research group on semiotics for information
systems. He has created the research (specialist) award of Information Systems with
Semiotics, the rst of this type in the United Kingdom.
Dr Liu rst worked in the Commission for Integrated Survey, China State
Planning Committee and Chinese Academy of Sciences. As a programmer and later
systems analyst designer, he was involved in and led a number of projects of devel-
oping information systems for regional planning and development purposes.
Originally trained as a computer scientist in his university education in China, his
postgraduate and doctorate education has been shifted towards management and
business systems; both of them were received in the Netherlands. He is one of the
main contributors to an information systems methodology, MEASUR (Methods
for Eliciting, Analysis and Specifying Users Requirements).
Dr Lius work is found in various computing areas, such as information systems
methodologies, requirements studies, information systems engineering, human
computer communication and collaborative work.
SEMIOTICS IN INFORMATION
SYSTEMS ENGINEERING
KECHENG LIU
Staordshire University
iuniisuio n\ rui iiiss s\xoicari oi rui uxiviisir\ oi caxniioci
The Pitt Building, Trumpington Street, Cambridge, United Kingdom
caxniioci uxiviisir\ iiiss
The Edinburgh Building, Cambridge CB2 2RU, UK
40 West 20th Street, New York, NY 10011-4211, USA
477 Williamstown Road, Port Melbourne, VIC 3207, Australia
Ruiz de Alarcn 13, 28014 Madrid, Spain
Dock House, The Waterfront, Cape Town 8001, South Africa
http://www.cambridge.org
First published in printed format
ISBN 0-521-59335-2 hardback
ISBN 0-511-03916-6 eBook
Cambridge University Press 2004
2000
(Adobe Reader)
To Lily and Jimmy

Their love is my drive
Contents
Preface page xi
1 Introduction 1
1.1 Information and information systems 1
1.2 Problems and challenges in information systems 2
1.3 Approaches and methods for information systems
development 5
1.4 MEASUR: a semiotic approach to information systems 7
1.5 About this book 8
Part one Semiotic framework and methods 11
2 Understanding semiotics 13
2.1 Signs and their functions 13
2.2 Semiosis and learning 15
2.3 Semiotics in computing 17
2.4 Semiotics in organisations and information systems 19
3 A semiotic framework for information systems 21
3.1 Philosophical stance 21
3.1.1 Objectivist paradigm 21
3.1.2 Subjectivist paradigm 24
3.1.3 Radical subjectivist paradigm 26
3.2 The semiotic framework 26
3.2.1 Physics 27
3.2.2 Empirics 28
3.2.3 Syntactics 29
3.2.4 Semantics 30
3.2.5 Pragmatics 31
3.2.6 The social level 33
3.3 An example of semiotic analysis 35
4 A semiotic approach to information systems development 37
vii
4.1 MEASUR 37
4.2 How MEASUR can help in information systems
development 39
4.2.1 Infrastructure analysis 40
4.2.2 Systems analysis, design and implementation 46
4.3 Summary 47
5 Knowledge representation and information analysis 49
5.1 Some basic considerations in knowledge representation 50
5.1.1 Expressive adequacy and notional eciency 50
5.1.2 Semantic primitives 50
5.1.3 Types of knowledge 51
5.2 Representation approaches 52
5.2.1 Typical examples 52
5.2.2 Conceptual graphs 54
5.3 Some fundamental issues of information analysis 56
5.4 The role of information analysis 58
6 Semantic Analysis 61
6.1 Theoretical aspects of Semantic Analysis 61
6.1.1 Aordances 61
6.1.2 Ontology and some other fundamental notions 63
6.2 NORMA 64
6.2.1 Well-formed formula 65
6.2.2 Aordance and ontological dependency 65
6.2.3 Semiotic behaviour 67
6.2.4 Time 67
6.2.5 Determiner and identity 68
6.2.6 Genericspecic relationship 68
6.2.7 Dening authority and responsibility 69
6.2.8 Graphic representation ontology chart 69
6.3 Using LEGOL to specify Norms 71
6.4 Conducting a Semantic Analysis 73
6.4.1 Understand the problem domain 74
6.4.2 Generating candidate aordances 75
6.4.3 Candidate grouping 76
6.4.4 Ontology charting 78
6.4.5 Norm Analysis 79
6.5 Commentary on Semantic Analysis 80
7 Pragmatics and communication 82
7.1 Human communication 82
viii Contents
7.2 Other approaches to communication 83
7.2.1 Speech Act Theory 83
7.2.2 Functional approach 86
7.2.3 Deontic logic for communication 89
7.3 Pragmatic aspect of human communication 94
7.4 The Norm Analysis method 98
7.4.1 The concept of norms 98
7.4.2 Norms in business organisations 100
7.4.3 Norm Analysis 102
7.4.4 Norms in computer systems 106
8 The social layer: modelling organisations as information
systems 108
8.1 Organisations as information systems 109
8.2 The notion of responsibility 111
8.3 An organisational morphology 112
8.4 Modelling the organisation 113
8.5 Summary: requirements for an eective information
modelling method 114
Part two Applications 117
9 From semiotic analysis to systems design 119
9.1 The semantic aspect of databases 119
9.2 Capturing the semantic aspect 120
9.3 Capturing the time aspect 121
9.4 Ontological modelling for conceptualisation 123
9.5 Intentions, propositional attitudes and consequent
operations 124
9.6 Other aspects of databases: facts, beliefs, and
knowledge 125
10 Semantic temporal databases 133
10.1 Databases 133
10.1.1 Developments in database management systems 133
10.1.2 Semantic temporal databases 136
10.2 The semantic templates 138
10.2.1 Dening a semantic template 138
10.2.2 ST for database design 139
10.3 Systems construction 142
10.4 LEGOL 143
10.4.1 Basic syntactic structure 143
10.4.2 Some important operations 145
Contents ix
11 Normbase: a new approach to information management 150
11.1 The Normbase concept 150
11.2 The Normbase system 152
11.2.1 The Normbase engine 152
11.2.2 The semantic temporal database 154
11.2.3 The norm store 155
11.3 Information management with the Normbase system 156
11.4 Using semiotic methods with other approaches 157
11.4.1 Relational database for implementation 158
11.4.2 Object-oriented methods for design and
implementation 159
12 Case study: development of a land resources information
system 164
12.1 Background 164
12.2 Semantic Analysis for requirements modelling 165
12.3 Norm Analysis 171
12.4 System design and implementation in the Normbase
approach 174
12.5 Discussions and conclusions 179
13 Case study: development of a test construction system 180
13.1 Background 180
13.1.1 CONTEST project 180
13.1.2 User requirements 181
13.1.3 Why choose Semantic Analysis? 184
13.2 System analysis 186
13.3 System design 190
13.4 System construction 194
13.5 Discussion and conclusions 194
Appendix A Semantic templates and surrogate specication 196
A.1 Denition of ST 196
A.2 Examples of using ST in discourse modelling 197
A.3 Examples of surrogates 199
Appendix B LEGOL applications in the CRIS case 201
B.1 Questions and LEGOL statements 201
B.2 Output from the Normbase 203
Bibliography 208
Index 217
x Contents
Preface
Information systems are a multi-disciplinary subject, whose objects of
study are information and its functions, information technology and its use
in organisational contexts. For nearly three decades, scholars and practi-
tioners have been pursuing eective paradigms, approaches, methods and
techniques for developing and engineering information systems. The book
is intended to contribute to this direction.
The research work on which the book is based began in 1989 in Twente
University, the Netherlands, where I joined a team led by Professor Ronald
Stamper. The team has been preoccupied by a series of philosophical and
methodological investigations into information systems for a long time. My
work at that time, with a focus on information modelling, was just a part of
the large programme entitled MEASUR.
The research programme began in 1973, marked by Stampers book on
information. It was rst called LEGOL, which aimed to deliver a set of
legally oriented techniques for requirements specication. It soon extended
into a research eort into Methods for Eliciting, Analysing and Specifying
Users Requirements (hence MEASUR). In the last ten years, the pro-
gramme has further expanded into a set of methods to deal with all aspects
of information systems. The theory of organisational semiotics is a key
foundation for the methods and techniques developed within MEASUR.
These methods and techniques enable one to understand and articulate the
business problem and its context under the study, to capture semantics and
intentions of users in requirement models, and also to implement technical
information systems that are exible and adaptable to the organisational
change.
This book focuses on the requirements engineering and development of
IT-based systems. After introducing basic principles of semiotics and the
relevance to information systems, the book presents the methods for
xi
requirements analysis and modelling, and techniques for implementing
technical systems. Finally, the last part of the book demonstrates the appli-
cation of these methods and techniques through a number of case studies.
I am deeply indebted to Ronald Stamper, my mentor and a true hero,
who has introduced me to this fascinating semiotic perspective of informa-
tion systems. Thanks are also due to Peter Andersen and Rodney Clarke,
with whom I have had inspiring discussions in the last few years on a
number of occasions.
Since I joined Staordshire University in 1993, I have been working with
a team of colleagues in a number of projects, including knowledge sharing,
information infrastructure, collaborative and software agent-based infor-
mation systems and electronic commerce. One particular project is on
requirements recovery in legacy systems re-engineering. I am grateful to my
colleagues in this team with whom I have had many opportunities to discuss
the ideas and apply these methods to various research problems: Albert
Alderson, Alan Dix, Hanifa Shah, Bernadette Sharp, Dave Brunskill, Sue
Blakey and Geo Crum.
A research-based masters course of Information Systems and Semiotics,
the rst of this kind, was launched two years ago at Staord. I must
acknowledge the students on this course and my other PhD students for
their enthusiasm in their exploring the relevance and power of semiotic
methods to their research work. Their feedback has been highly valued.
I would like to thank David Tranah of CUP, and Peter Jackson, my copy
editor, for making this book possible; particularly for Peters thorough and
rigorous editing, which has made this book a smooth and pleasant read!
KL
Staord
xii Preface
1
Introduction
1.1 Information and information systems
What is information? Many people have attempted to give a denition but
most of them are not complete. A typical explanation is that information is
processed data that has meanings to its users. But then questions arise in
what meaning is. If information is to the study of information as object is to
physics, and there are many laws by which we can study objects, then what
are the laws by which we can study information? What is the study of infor-
mation anyway?
What can be said here is that information is not a simple, primitive
notion. Devlin (1991) compares the diculties for a man in the Iron Age to
answer the question What is iron? and for a man in todays Information
Age the question What is information? To point to various artefacts of
iron in order to answer his question would not be satisfactory; to demon-
strate some properties of information as an answer to What is informa-
tion? is not good enough either. People can feel the possession of
information, and can create and can use information. They gather it, store
it, process it, transmit it, use it, sell it and buy it. It seems our lives depend
on it; yet no one can tell what exactly it is.
In order to understand the nature of information, one may have to nd
some fundamental and primitive notions with which the question can be
investigated and explained. The concept of a sign is such a primitive notion
that serves the purpose. All information is carried by signs of one kind or
another. Information processing and communication in an organisation
are realised by creating, passing and utilising signs. Therefore, understand-
ing signs should contribute to our understanding of information and infor-
mation systems (Stamper 1992).
The investigation of information systems is an active area where much
attention has been paid by other research and industrial communities.
1
Much discussion has taken place on its pluralistic and interdisciplinary
nature and its foundations. An increasing number of researchers and prac-
titioners dene information systems as social interaction systems. Social
and organisational infrastructure, human activities and business processes
are considered as part of information systems. Information systems in this
denition can produce messages, communicate, create information, and
dene and alter meanings. The UK Academy of Information Systems pro-
vides the following denition of the domain of information systems
(UKAIS 1996). The study of information systems and their development
is a multi-disciplinary subject and addresses the range of strategic, manage-
rial and operational activities involved in the gathering, processing, storing,
distributing and use of information, and its associated technologies, in
society and organisations. Many authors emphasise the importance of
information systems study in this postmodern society and its interdiscipli-
nary nature. For example, Avison and Nandhakumar (1995) suggest that
information systems encompass a wide range of disciplinary areas such as
information theory, semiotics, organisation theory and sociology, com-
puter science, engineering, and perhaps more.
1.2 Problems and challenges in information systems
Information processing has become a major industry which plays a
signicant role in a national economy (Machlup 1980). Porat (1977), based
on analysis of various sources, suggests that in developed countries such as
the USA and west European countries, about 50 per cent or even more of
the labour force is engaged in information work. However, the following
occurrences have been reported in numerous studies: large company
budgets for IT (information technology) made, huge expenses on systems
development projects spent, and, yet, low economic return from the high
investment. Strassmann (1990) argues that there is no guarantee that large
investment on information technology will lead to high business perfor-
mance or an increase in revenue: no relationship between expenses for
computers and business protability on any usual accounting basis across
an industry. Some companies use IT well but others balance this positive
eect with poor returns on IT investment. Many surveys in industry exhibit
evidence that a considerable number of information systems developed at
great cost fail to satisfy users or have to be modied before they become
acceptable. According to an analysis done by the US General Accounting
Oce in the late 1980s, among a group of US federal software projects
totalling $6.2 million, less than 2 per cent of software products were used as
2 Introduction
delivered, more than half were not used (including those not delivered).
The situation has not been improved. In 1993, the Taurus project (Transfer
and Automated Registration of Uncertied Stock) of the London Stock
Exchange was aborted after the expenditure of 400 million over the pre-
ceding ve years (De Volkskrant, March 12, 1993; Computing, March 15,
1993). In 1997, Advanced Technical Strategy Inc. published their study
The software development crisis on the Internet, as quoted in Figure 1.1
(Boustred 1997). In this gure, the shameful numbers show a majority of
projects cost nearly twice their original budgets. The customer acceptance
gures display that more than two-thirds of the software products were
never used or never completed. One of the main reasons for these failures is
inappropriate users requirements studies which lead to incorrect systems
analysis and design.
The social, cultural and organisational aspects play more decisive roles
than technology itself. A computer is worth only what it can fetch in an
auction, Franke (1987) writes. It has value only if surrounded by appropri-
ate policy, strategy, methods for monitoring results, project control, talented
and committed people, sound relationships and well designed information
systems. These information systems are composed of organisational infra-
structure, business processes and technological systems. An organisation
with an improperly designed bureaucracy and unprepared culture will not
guarantee that the possession of more information will improve the quality
of management; on the contrary, the quality of management will suer
from being overloaded with irrelevant information (see Acko (1967)). As
Lee (1988) points out, . . . certainly, this [information] technology has done
much to improve bureaucratic eciency. On the other hand, we claim, it has
not helped at all in the management of bureaucratic complexity. Indeed, it
has aggravated the problemby obscuring bureaucratic rules and procedures
1.2 Problems and challenges in information systems 3
Shameful numbers
31.1% of projects are cancelled before they ever get completed
52.7% of projects go over time and/or over budget, at an average cost 89%
of their original estimates
16.2% of software projects are completed on time and on budget
In larger companies only 9% of their projects come in on time and on
budget with approximately 42% of the originally proposed features
and functions
In small companies 78.4% of their software projects get deployed,
with at least 74.2% of their original features and functions
Figure 1.1. Shameful numbers in software development crisis.
in the form of computer codes. (For instance, many have experienced the
frustrations of trying to rectify a computer-generated billing error).
There are two other problems observed in systems developments. One is
the long lead-time between the commencement of the projects and the
systems becoming available. The commencement stage may be to study
business strategy of a given organisation, to conduct information systems
planning, followed by business analysis, system design, construction
design, etc. (see the stages considered in the information systems method-
ologies framework discussed in Olle et al. 1991). Many systems develop-
ment projects have taken too long in reaching fruition. Hence the
immediate risk arises that, during the development processes, the users may
wish to change the requirements they specied when the projects began. If
the development methodologies do not allow the incorporation of the
modied requirements into the developments, then the products after the
long lead-time may hardly satisfy the users present needs.
Another problem observed in practice is related to systems analysis, par-
ticularly to analysis documentation. Applications of most development
methodologies produce huge volumes of documentation. These documents
are supposed to address the technical people involved in the development
projects. Because the primary purpose of the analysis documents is usually
to function as a formal basis for the next stage, i.e. systems design, the deliv-
erables from analyses are often directed particularly to systems designers in
the development teams. The business users, who may actually be the owners
of the systems, are not considered as recipients of the information analysis
products. As a result, the analysis products are presented in some articial
languages that are specially invented by and for the systems development
communities. Syntactic elements such as jargonised notation (of the
method of analysis used) and technical disciplines (e.g. logical consistency
and data normalisation) receive more attention in those methodologies
than semantic aspects. The large volume of documentation may contain a
great deal of design symbols and special notation. Each project has to
dene the meanings of the terms used in the information system in a data
dictionary. The products of analysis may then be very precise and conve-
nient for the next step, i.e. systems design; they may even be automatically
translated into a database schema by a CASE (Computer Added Software
Engineering) tool. These documents are too dicult for the systems
owners to understand.
The users need to check above all the meanings incorporated into the
specication in order that results of the analysis can be veried. However,
4 Introduction
documentation in most of the conventional methodologies creates barriers
to, rather than facilities for, requirements analysis. Voluminous documen-
tation builds up psychological diculties; in addition to that, many compli-
cated technicalities have to be learned through undergoing intensive
training. Every time the users want to understand a fragment of the repre-
sentation, the data dictionaries have to be used. Even with the help of data
dictionaries, the users are still unlikely to be able to make an adequate check
on the meanings of words. The risk arising from such a situation is that the
users requirements may be wrongly understood by the analysts. Wrong
requirements may be sought, though represented in a correct syntax, as
foundations for systems development. In such cases, it is unwise to expect
satisfactory information systems to be produced based on these require-
ment models. As a solution, it is suggested that a method with an emphasis
on semantics is needed. Possibilities of preserving and clarifying meanings
in requirements analysis should be considered as more important criteria
than any others. Therefore, a guarantee of producing a correct requirement
analysis is possible only if the users can validate and verify an information
systems model, which can be technically implemented. Ideally, this model
should be a by-product of the business analysis.
1.3 Approaches and methods for information systems development
Frameworks and methodologies have always been regarded as fundamen-
tals for information systems. Many authors oer comprehensive reviews
and critique of frameworks and methodologies for information systems
(see, for example, Olle et al. (1991), Hirschheim et al. (1995), Avison &
Fitzgerald (1995)). A suggestion to characterise the methods is made, with
three features: data-oriented, process-oriented and behaviour-oriented. An
elaborate analysis and comparison of some methodologies and models for
information systems can be found in Hirschheim et al. (1995) where
methodologies and models are evaluated against some 50 criteria which are
grouped into technical, usage, economic and behavioural aspects. It is sug-
gested that there are four paradigms: functionalism, social relativism,
radical structuralism, and neohumanism (Hirschheim & Klein 1989;
Hirschheimet al. 1995). Each information systems development approach
may be based on one of the paradigms. Goguen (1992) discovers two cul-
tures in information systems development and requirement engineering:
the dry and the wet. Bickerton and Siddiqi (1993) classify some 30
methods for systems development and requirements engineering in a
1.3 Approaches and methods 5
framework in which two of the important considerations are whether the
methods are hard or soft in character. More studies and comparisons
of information systems development methodologies can be found in the
literature.
There are confusions in concepts and theories of information systems,
which actually contribute to the diculties of development and selection of
a suitable methodology. The FRISCO task group (FRamework of
Information System COncepts) of IFIP WG 8.1
1
has set up as its goal the
removal of the confusions in concepts and theories (Falkenberg et al. 1998).
The following passage (quoted from Lindgreen (1990)) manifests dissatis-
faction about information systems development:
There is a growing concern within IFIP WG 8.1 about the present situation, where
too many fuzzy or ill-dened concepts are used in the information systems area.
Scientic as well as practice-related communication is severely distorted and ham-
pered, due to this fuzziness and due to the frequent situation that dierent commu-
nication partners associate dierent meanings with one and the same term. There is
no commonly accepted conceptual reference and terminology, to be applied for
dening or explaining existing or new concepts for information systems.
Agreat deal of eort has been put into information systems studies. One of
the purposes is that criticism and suggestions regarding the methods can
be learned from, and improvements can be made. There are some problems
noticed in many of the above mentioned studies. One of the problems is
that the majority of the commercially available methods for information
systems development are dry or hard in nature; social and organisa-
tional aspects tend to be ignored. However, as noted in the preface of
Galliers (1987), Few would argue that the study of information and the
need to treat the development of information systems from the perspec-
tives of social as opposed to technical systems remains as crucial today as it
was when Stamper rst published his book in 1973.
2
Motivated by these
same reasons, many people have put eort into bringing about workable
theories and methods for information systems development enabling
people to handle both the social and technical systems (see, for example,
Mumford & Weir (1979), Checkland (1981), Lyytinen & Lehtinen (1986),
Stowell (1995)).
6 Introduction
1
WG 8.1, part of IFIP (the International Federation for Information Processing), is one of
the Working Groups on Information Systems, and focuses on methodologies and issues of
design and evaluation of information systems.
2
Stamper (1973) began his book with a signicant remark that the explosive growth of infor-
mation technology has not been accompanied by a commensurate improvement in the
understanding of information. He appeals for an understanding of both machine and
human information systems.
1.4 MEASUR: a semiotic approach to information systems
MEASUR (Methods for Eliciting, Analysing and Specifying Users
Requirements) is a research programme initiated in the later 1970s by
Ronald Stamper. The main objective of the programme is to investigate and
deliver a set of methods that can be used by researchers and business users
in their understanding, development, management and use of information
systems. The research programme has evolved in the last two decades. As he
later elaborated (Stamper 1993), MEASUR is now a rich acronym:
Methods, Means, Models . . . for
Exploring, Eliciting, Evaluating . . .
Articulating, Analysing, Assessing . . . and
Structuring, Specifying, Stimulating . . .
Users
Requirements
The various MEASUR methods enable one to start with a vague and
unstructured problem, perhaps in a messy problem situation, and gradually
dry it out until it is crisp and precise enough to derive a set of technical
solutions. MEASUR is focused upon solving business problems in the
broadest sense. It helps in solving a wide range of problems, especially
those that require organisational or social intervention to solve them.
MEASUR addresses information technology problems, as well as organi-
sational problems. In dealing with information technology problems, it can
handle the up-stream end of software engineering. In a conventional situ-
ation it would primarily serve the managers and other system users, helping
them to identify and solve their problems, and lead them to a precise state-
ment of information requirements. In this role it covers the domain of
information strategy and planning but it also provides detailed
specications of any problem domain where detailed requirements
specications are needed before designing a computer based system.
Further descriptions of MEASUR methods are found in Chapter 4. Here
an examination of the major philosophical assumptions will be given.
Stamper (1992) proposed a new paradigm for MEASUR: the informa-
tion eld. As opposed to information ow, which is the basis for most of the
conventional information systems approaches, this information eld para-
digm enables to us understand information from a new perspective and
therefore to develop information systems more properly. A physical
analogy can be used to illustrate this information eld paradigm (Huang
1998). Just imagine how a space vehicle in the sky is under the interacting
1.4 MEASUR: a semiotic approach 7
inuences of many dierent internal and external elds: elds of gravita-
tion, electro-magnetic forces as well as clouds of gas, and internal, elastic
tensions. There is no point trying to explain the vehicles behaviour in terms
of energy, momentum and materials exchanged between its components
we need a macro model. The information eld paradigm is the macro
model. It helps us to obtain a macro perspective before one works on details
of ows where that is possible.
MEASUR rejects the position taken by many practitioners in this eld in
that they tend to consider information systems as devices for representing
and interacting with some objective reality. It takes a dierent stance.
MEASUR is based on the assumption that the world is constructed socially
and subjectively. It recognises that in and out of a business system, there are
many actors, or human agents. The owners, managers, sta, suppliers,
clients, professional groups, local communities, and so on, are all governed
by the forces in the information elds and therefore behave accordingly.
These forces are related to their interests, assigned functions, tasks, objec-
tives, personal values and organisational goals. These forces may be present
in the forms of formal and informal rules, beliefs, cultural habits and con-
ventions, which can be called norms.
This subjective, organisational view of information systems and the
information eld paradigm lead to the MEASUR approach to information
systems work. In the development of an information system, particularly in
requirements analysis and representation, the scope of attention should be
the whole organisation instead of only the part of the business operations
that is going to be automated by the technical systems. The foci of analysis
should be actors (i.e. agents) and their behaviour which are governed by
social, cultural, institutional, economic and other kind of norms.
Therefore, an eective way of representing users requirement is to describe
the agents and their intended patterns of behaviour in terms of social
norms.
1.5 About this book
One of the motivations of this book is to examine the relevant issues in
information systems development. It will then introduce a semiotic
approach with a set of methods initially developed by the MEASUR
research team. This team has now grown to a community with colleagues in
many countries and the MEASUR methods have been applied in many
research and industrial cases. But little systematic account can be found.
8 Introduction
This book aims to ll this gap amongst many MEASUR paper publica-
tions, technical papers, research reports and project documentation.
The structure of the book is in two parts. Part one (Chapters 28) focuses
on discussion of semiotics, the semiotic framework for information
systems, and the semiotic methods for information systems analysis. Part
two of the book (Chapters 913) puts emphasis on the application of the
methods. It describes how the methods are used in systems analysis, design
and implementation of a computer information system.
Chapter 2 is about principles of semiotics and the relevance of semiotics
to organisations and computing. Chapter 3 examines dierent philosophi-
cal paradigms and then introduces a semiotic framework for information
systems. The semiotic framework represents a radical subjective view of the
social world, organisations and information systems. It serves as philo-
sophical and methodological guidance for the development and formula-
tion of semiotic methods to be discussed later. In Chapter 4, an overview of
the semiotic methods for information systems, MEASUR, is presented.
MEASUR comprises a set of methods covering all stages and aspects of
planning, developing and maintaining an information system. Although
the rest of the book focuses on the tasks of systems analysis and design,
with a detailed account of the methods of Semantic Analysis and Norm
Analysis, the chapter also discusses the use of other MEASUR methods.
From Chapter 5 to Chapter 8, aspects in three layers of the semiotic
framework are followed: semantic, pragmatic and social aspects. The reason
for such a careful investigation of these aspects is that they are more rele-
vant than another three (physical, empirical and syntactical) aspects as far
as information systems analysis and design are concerned. For example,
physical and empirical issues can be important in designing and implement-
ing a computer network, and syntactical issues can be highly relevant in
programming and implementing a database. The tasks of systems analysis
and design are primarily concerned with capturing requirements, studying
business knowledge and representing knowledge at semantic, pragmatic
and social levels. Before the semiotic methods are discussed, Chapter 5
explores the basic notions of knowledge representation and information
analysis, and presents basic considerations of those methods. When one is
equipped with fundamentals of knowledge representation and systems
analysis, Chapter 6 begins to tackle the issues at the semantic level. The
method of Semantic Analysis, which serves as a core of other methods dis-
cussed later, is described here. Chapter 7 tackles the next layer of the semi-
otic framework, pragmatics and communication. It also examines other
1.5 About this book 9
approaches before looking at Norm Analysis. Finally in this part, we come
to the social layer in Chapter 8, which any technical system should ulti-
mately serve. It is argued that an eective modelling method should enable
one to analyse all these issues and it demonstrates that these semiotic
methods are capable of this.
The second part of the book (Chapters 913) deals with the use of
methods in systems development. Chapter 9 displays the performance of
analysis to capture semantic and pragmatic information in information
model. Chapter 10 shows how semantic and temporal information can be
stored in databases, and processed and used with the help of a semantic
temporal database language, LEGOL. Chapter 11 exhibits the application
of the semantic approach to the entire development process of information
systems. In addition, hybrid approaches with input from the semiotic
methods are also discussed. Chapters 12 and 13 contain two case studies to
illustrate the methods and their applications in systems development proj-
ects.
The book also contains two appendices. Appendix A gives a full account
of the semantic temporal databases, which helps one to understand and
appreciate better the power of such databases. Appendix B presents the
CRIS case and illustrates the use of LEGOL.
10 Introduction
Part one
Semiotic framework and methods
2
Understanding semiotics
The word semiotics comes from the Greek for symptom. Charles
Sanders Peirce (18391914), who was primarily a logician and competent in
mathematics and many other branches of science, including astronomy,
chemistry, physics, geology and meteorology, founded semiotics as the
formal doctrine of signs. At almost the same time, Ferdinand de Saussure
(18571913) founded semiology, a European school of semiotics.
Semiology aims at discovering what constitutes signs and what laws govern
them (a quote of Saussure, cited in Hawkes (1977, p. 123)), with a slight
emphasis on social psychology and general psychology. Semiotics covers
the whole cycle of a sign from its creation, through its processing, to its use,
with more emphasis on the eect of signs. Although, in the literature, some-
times no clear distinction is made between semiotics and semiology, it is the
former which will be the theme of study in this book.
There are three distinct elds of semiotics: syntactics (or syntax), seman-
tics and pragmatics, which go back ultimately to Peirce (cf. Lyons (1977)).
Morris (1938) made semiotics more generally familiar as a science of signs,
to which he made many important contributions, largely from a behav-
ioural standpoint. According to Morris, pragmatics deals with the origin,
uses and eects of signs within the behaviour in which they occur.
Semantics deals with the signication of signs in all modes of signifying;
syntax deals with the combination of signs without regard to their specic
signication or their relation to the behaviour in which they occur.
2.1 Signs and their functions
The notion of sign is essential in semiotics. A sign is something which
stands to someone for something else in some respect or capacity. Every
sign involves a signier the material form of the sign and the signied
13
the object, action, event or concept it represents. Note that there is always
someone an interpretant involved, to whom the signication makes
sense. An example is given below to illustrate this.
Sign The written word house or a drawing of a house.
Signied The category house.
Interpretant Anyone who is involved in reading and interpreting the sign.
The relationships between the signier and signied may not be xed and
may dier depending on the context, culture and language. For example the
word slim, although the spelling is the same in both languages, means
cleaver in Dutch while in English it means thin.
Peirce named three categories of signs: icon, index and symbol. An icon
is a sign which resembles the signied. An iconic sign represents its object
mainly by its similarity and signies by virtue of metaphor. Icons could also
be verbal by making words onomatopoeic. Examples for iconic signs are
images, diagrams, maps, portraits, photographs and icons of the algebraic
kind. An indexical sign is inherently connected in some way to the signied
object. An index is normally instructive and signies by virtue of
metonymy. Examples are smoke signifying re, a thermometer, a knock on
a door, footprints, and pain in the stomach. The third type of sign is a
symbol whose relationship with its signied object is arbitrary or purely
conventional. It signies by virtue of an arbitrary norm which may be
rooted in societal or cultural convention; for example, the three colours of
trac lights. The three categories of signs are not separate and distinct; a
complex sign may be a mixture of several kinds.
A primary purpose for human beings to create and to use signs is com-
munication, unless a sign is purely for private use. To be able to use signs
properly, one needs to understand the relationships between the signs and
the things they refer to (i.e. referents). For the three categories of signs, the
degrees of connection between the signier and the signied are dierent.
Indexical signs have existential or causal relationships with the signied.
Iconic signs such as photographs are physically forced to correspond to the
objects they represent. However, as discussed by Fiske (1990), there are
varying degrees of arbitrary elements with an iconic sign. A portrait is less
iconic or more arbitrary than a photograph; a cartoon is more arbitrary
than a portrait. But even with a photograph, the cameraman can take the
picture of dierent parts of an object, from dierent angles and with
dierent lighting, to incorporate some arbitrary elements. Therefore con-
vention, cultural norms in a society, is necessary to understanding of any
sign, however iconic or indexical it is (Fiske 1990, p. 53). Norms play an
14 Understanding semiotics
important variety of roles in communication and signication. They can be
as formal as the rules to dene the relationship between signs and the
signied, and to govern how the signs should be used. Think about English
grammar in forming singular and plural nouns. There are probably more
informal norms that one can learn from experience, for example, the
meaning of zooming into some football action and playing it in slow
motion on television (a highlight or a mistake?), and the meaning of a
sudden start of music in a romantic lm. Convention is the social dimen-
sion of signs. It is the agreement amongst the users about the appropriate
uses of and responses to a sign. One has to understand and to follow social
and cultural norms when a sign is created and used; otherwise it will be only
private and thus does not function in terms of communication.
2.2 Semiosis and learning
The subjects of study for semiotics are all kinds of signs: verbal language,
pictures, literature, motion pictures, theatre, body language, and more. The
central concept of semiotics is semiosis, which was introduced by Morris.
Semiosis is a process of understanding involving something as a sign to
some organism. In his book, Nauta (1972) describes semiosis as a sign-
process exercise of mediated-taking-account-of. The mediators are sign
vehicles; the taking-account-of is interpreters; what is taken account of is
designata. These four elements function in every sign-process or semiosis.
The sign vehicle is the something which is a sign to an organism (i.e. the
interpreter). Sign vehicle is dened as a particular event or object, such as a
soundor mark, that functions as a sign. Thus a signvehicle is a special kind
of stimulus. In its function as a sign, the sign vehicle refers to something, the
denotatum; it also has a special eect on the attitude of the interpreter; this
alteration of the inner state of the interpreter is called the interpretant.
Figure 2.1 illustrates the basic concepts involved in a semiosis and the
relationships between these concepts. A semiosis is the sign-mediated
process in which a knowing subject gives meaning to what is taken to be an
event or actuality. Formulated in terms of the three universal categories, it
is the universal process in which a rst, a presentation (which could be any
quality, any thing or any idea) functions as a representamen or sign in being
applied to a second. In that, a second is an object or actuality, through the
intermediary of a meaning-mediation third (Kolkman 1993). The notion of
interpretant is richer and broader than interpretation. It also includes
action and feeling. An interpretant can in turn be treated as a sign in
another process of semiosis.
2.2 Semiosis and learning 15
There are four characteristics for describing semiosis. Firstly, it is univer-
sal. It is applicable to any type of sign-processing activities. It explains the
mechanism for both creating and using a sign. Secondly, semiosis is a
process capable of identifying anything present according to a specic crite-
rion or a norm. The process of representation can be recursive: a sign can
be seen as an object in another sign-process, just as an interpretant or an
object can be a sign. This will lead to its fourth characteristic which is the
possibility of making anything not present identiable. Finally, semiosis is
subject-dependent. It is closely related to the interpretant who can be an
individual, a group or a social community with certain knowledge and
obeying certain norms. A semiosis process is always partial in representa-
tion of the object and subjective in interpretation, depending on the view-
point of the interpretant, and the knowledge and ability the interpretant
possesses.
Semiosis can be seen as the building of structures of experience via signs.
Thinking will involve using signs and inferring an object from its sign. That
inference, the eect of the sign, is the interpretant (Cunningham 1992). The
traditional account of inference usually suggests two processes: deduction
and induction. However, Peirce proposed three irreducible, distinct modes
of inference in semiosis for understanding and learning. The additional one
in the Peircean logic system is abduction. The logic of abduction (the
rstness) and deduction (the secondness) contribute to our qualitative or
conceptual understanding of phenomena, while the logic of induction (the
thirdness) adds quantitative details to the qualitative or conceptual knowl-
edge (Yu 1994).
THIRDNESS
interpretant
e.g. concept HOUSE
FIRSTNESS
sign
e.g. word house
SECONDNESS
object/referent
e.g. concept house
d
e
c
o
d
i
n
g
e
n
c
o
d
i
n
g
i
n
d
i
c
a
t
i
o
n
d
e
n
o
t
a
t
i
o
n
I
S O
Figure 2.1. Semiosis as a process of semiotic activity.
The three types of inference are explained below with examples to illus-
trate the dierences between them. In the case of deduction one tries to
make a conclusion or to nd a result on the basis of an observed fact by
application of a rule.
Deduction: Rule : All the apples from the bag are red
Fact : These apples are from the bag
Result/conclusion: These apples are red.
Induction is related to the generation of a general rule on the basis of a fact
and conclusion.
Induction: Fact : These apples are from the bag
Result/conclusion: These apples are red
Rule : All the apples from the bag are red.
Abduction in Peirces theory is the process of forming a hypothesis that
explains the given observations (see Peirce (193158)):
The surprising fact, C, is observed;
But if A were true, C would be a matter of course,
Hence, there is reason to suspect that A is true.
As Yu (1994) summarises, deduction draws logical consequences from
premises. This kind of reasoning does not lead to the discovery of new
knowledge, because the conclusion has already been embedded in the
premise (Peirce [1900] 193158). Inductive reasoning is based on generality
and the law of large numbers of cases. Induction generates empirical laws
but not theoretical laws, because the conclusion one reaches from research
with a large number of samples is merely probably (never certainly) true.
Finally, abduction is to look for a pattern in a phenomenon and suggest a
hypothesis. The objective of abduction is to generate hypotheses and,
further, to determine which hypothesis or proposition to test, not which one
to adopt or assert. To achieve a comprehensive inquiry, one should apply
abduction, deduction and induction all together. Abduction and deduction
are the conceptual understanding of a phenomenon, and induction is the
quantitative verication. In short, abduction creates, deduction explicates,
and induction veries (Yu 1994).
2.3 Semiotics in computing
Semiotics has been an area actively attended to by scientists in linguistics,
media studies, educational science, anthropology, and philosophy of lan-
2.3 Semiotics in computing 17
guage. Semiotics as a whole or some of its branches have inuenced many
of these studies. Computing, including the development and use of com-
puter systems, is another eld in which semiotics has shown a great rele-
vance.
Computer semiotics, as dened by Peter Andersen (1991), studies the
special nature of computer-based signs and how they function. It is an
emerging discipline that will guide people in design and development of
computer systems. In Andersens map of a computer semiotics (Andersen
1990 Figure 2.2), the perspective of viewing the signs as systems occupies
the centre. In addition there are three main branches: the psychological
view of signs as knowledge, the sociological view of signs as behaviour, and
the aesthetic view of signs as artefacts. In the signs-as-systems perspective,
the individual is considered as a creator, interpreter and referent of signs, or
as a user who makes use of a semiotic product from others. In the signs-as-
knowledge perspective, the individual is considered as an assemblage of
parts: the persons biological ability and psychological mechanism will
enable them to learn; so the use of signs is a processing of understanding
and learning. Signs, particularly languages, can serve as a means of describ-
ing, expressing, designing and constructing both social and computer-
based systems. Winograd & Flores (1987) also advocate the same approach
and illustrate how information systems development can be started from
social reality as the natural foundation.
Applications of semiotics and its methods and techniques can be found
in many areas of computing such as computational linguistics, information
Program development
Interface design
Aesthetics
Signs as
art(efacts)
Signs as
system
System
description
Systems
development
Signs as
behaviour
Signs as
knowledge
Computer supported
collaborative work
Work analysis
Organisational analysis
Technology assessment
Cognitive science
Cognitive ergonomics
Figure 2.2. Map of a computer semiotics (Andersen 1990, p. 18).
systems, computer-supported collaborative work, knowledge engineering
and articial intelligence (see, e.g., Benyon (1994), Cunningham (1992),
Fiske (1990), Gonzalez (1997), Klein (1996), Liu & Gao (1997), Liu & Dix
(1997)).
2.4 Semiotics in organisations and information systems
Organisational semiotics is one of the branches of semiotics particularly
related to business and organisations. As manifested in the workshop which
was attended by many working in various application areas of semiotics
(OSW 1995), organisational semiotics is the study of organisations using the
concepts and methods of semiotics. This study is based on the fundamental
observation that all organised behaviour is eected through the communi-
cation and interpretation of signs by people, individually and in groups.
The functions of creating, reproducing, transmitting, transforming, pre-
serving and manipulating signs have always been performed with the aid of
various technologies. Semiotics, as the doctrine of sign, draws from many
disciplines to throw light on the ways in which people use signs for all kinds
of purposes. The scope of organisational semiotics includes both public
and private organisations and concerns their inner workings, their inter-
actions with the environment and with one another. The aims of this study
are to nd new and insightful ways of analysing, describing and explaining
organisational structure and behaviour. The results of this work will in
valuable ways relate to and augment current theories about organisations
that are based in many disciplines such as economics, social theory, social
psychology, anthropology, history, law, information systems, and so on.
Specic problem domains being addressed by people working in organi-
sational semiotics include aspects of such elds as marketing, human
resource management, business law and business ethics, the historical
development and future evolution of the organs of government, religion,
politics, justice, social and health care, as well as private and public busi-
nesses. The study is not conned to information expressed in speech,
writing or graphics, but takes account of the semiological aspects of the
products and productive resources of organisations including the architec-
tural forms which they inhabit. It is relevant to the conservation and the
changing of organisational structure and behaviour. Although the subject
is wide and touches many other specialist branches of semiotics (e.g. educa-
tion, law, architecture, media and so on), it only overlaps with them where
they are concerned with matters of organisation in their areas.
Semiotics has also been applied to information systems work. During the
2.4 Semiotics in organisations and information systems 19
last decade information systems research has grown dramatically to
embrace the insights of several philosophical traditions and semiotics has
been one of those contributing to information systems research. It has
recognised that within the research community, there are at least two major
camps: realists and constructivists. For realists, recognising that the natural
language is the ultimate meta-language, their tasks are to devise eective,
formal languages as means of modelling the structure and behaviour of
organisations; and to improve the semantics preserved in the models.
Constructivists on the other hand consider that language is not only a
medium of expression but also an active force shaping the way the mind
functions. By using language, one is engaged in a social process of sense-
making and reality construction. Eorts have been made through the study
of signs and language to facilitate communication among adherents to
alternative philosophical traditions in information systems research (Klein
1996).
3
A semiotic framework for information
systems
A semiotics of Stamper (1973), based on the work of Peirce (193158),
Morris (1938, 1946) and on others, and on information theory (Shannon
& Weaver 1949), has evolved into a set of semiotic methods for information
systems. A radical, subjectivist stance has been accepted as the basic philos-
ophy for developing the set of methods and tools for information systems
development. In this chapter, the discussion will focus on the examination
of the subjectivist paradigm and the semiotic framework (Stamper et al.
1988).
3.1 Philosophical stance
A paradigm is a set of the most fundamental assumptions adopted by a
professional community that allows its members to share similar percep-
tions and engage in commonly shared practices. Scientic achievements
recognised in the particular community and research traditions educate its
members to commit themselves to the same rules and standards for
scientic practice (Kuhn 1962). Adoption of a certain paradigm will deter-
mine the way of exploring the world; it will determine the research theory
and method. The usual considerations on scientic paradigms may be com-
plemented by considering a set of presuppositions as paradigmatic, thus
relating the discussion to philosophy.
3.1.1 Objectivist paradigm
A predominant philosophical position is the objectivist paradigm which
assumes that there is a subject-independent world, some objective reality.
Objectivism in its rigorous forms regards the universe as being made up of
21
self-existent entities. The main tenets of objectivism are threefold (Mishra
1982):
The entities (facts, objects) under study in logic, mathematics and the
physical sciences are not mental in any usual or proper meaning of the
word mental.
The existence and nature of these entities are not in any sense qualied by
their capacity for being known.
The degree of unity, consistency or relatedness subsisting among entities
is a matter to be empirically ascertained.
In this stance the human mind is treated as a blank sheet on which all the
knowledge that it later comes to acquire is written by the external world.
Individuals are passively receiving similar ideas. They will arrive at a
natural consensus, which is dictated by the external world. Because human
perceptions are seen as a kind of mirror image of the world in the mind, it is
believed in this paradigm that if one makes the investigation carefully
enough one will arrive at the same picture of the world as everyone else.
Natural sciences presuppose that there is a body of knowledge which is uni-
versally true. The task of scientists is to continuously discover the unknown
part of the world and to perfect the knowledge. Sucient education will
enable one to know the right way of learning and to be able to deduce new
knowledge, which should reach the same conclusion as that of anyone else,
based on what is known to him.
These basic assumptions may be justied in some limited circumstances
if the area of study is of a mechanical nature, regardless of the human
impact. The objectivist position urges one to apply models and methods
derived from the natural sciences to the study of human aairs. The social
world is treated as if it were some natural world composed of billiard balls
(see, for example, Burrell & Morgan (1979), Hirschheim & Klein (1989)).
However, a quick observation of day-to-day activities will tell us a dierent
story. In our social aairs, there are always dierent opinions involved. The
notion of true universal knowledge seems not to work as it is supposed to
by the objectivist. Dierences in political and economic interest, and cul-
tural background, lead to diversied judgements instead of a common con-
clusion on the basis of the objective truth. To reach a consensus, for
example, about a management decision, normally requires a lot of eort in
discussion, clarication and negotiation, though members of the manage-
ment team live in the same physical world. This is because of the subjectiv-
ity of knowledge, experience and perception.
This objectivist position leads to a set of basic beliefs in the elds of
22 A semiotic framework for information systems
language and language use, communication, and information systems.
Hirschheim & Klein (1989) illustrated that there are dierent paradigms of
information systems development which are named as functionalism,
radical structuralism, social relativism, and neohumanism; they can be
placed in a coordinate system (Figure 3.1).
In his criticism, Lyytinen (1987) characterises the objectivist position in
the systems analysis and development by examining the following four
objectivist principles.
(1) Correspondence There is a xed, immutable entity-structure in an enter-
prise. The description of the problem domain is a representation of
reality which has a correspondence between the words, and entities and
relationships. The problem of this principle arises when dealing with
soft entities, such as legal person, loan, customer, order, etc. A
great deal of disputes in legal cases was caused by ambiguous meaning
for such soft entities.
(2) Objectivity This principle suggests that the statements or observations
of a problem domain are independent of an observer, because observa-
tions by anyone should be the same. According to this principle, a sen-
tence like The marketing director believes that the sales have dropped
by 10% is meaningless, because someones belief suggests that it may
not be a fact, which is beyond the scope of discussion. The objectivist
position tends to ignore the contextual information associated with the
observations. Modalities, purposes and intentions are ltered out from
the statements.
(3) The excluded middle This principle is rooted in one of the fundamental
ORDER
OBJECTIVISM SUBJECTIVISM
CONFLICT
Functionalism
Social
relativism
Radical
structuralism
Neohumanism
Figure 3.1. Dierent paradigms for information systems studies (Hirschheim &
Klein 1989).
theories of formal logic, bivalence, which says that every sentence,
statement or proposition is either true or else false (Note: for more on
the denition of bivalence see the glossary of Haack (1978)). However,
social organisations are not built up on the basis of a set of axioms and
they do not always function following a nite number of logical rules.
Despite a general applicability of norms, there are exceptional situa-
tions where one is allowed to violate a rule (Holy & Stuchlik 1983; Lee
1988; Ryu & Lee 1993). For example, a police car in an emergency situ-
ation may disobey trac signals. Norms do provide standards of
conduct, but more often than not, deviance of behaviour occurs in
society (Gibbs 1981) due to dierences in standards and beliefs.
Because of these dierences, behaviour considered right or wrong
based on some standard may be judged neutral based on other stan-
dards.
(4) Language neutrality This assumes a descriptive role of language. A lan-
guage is seen as a means of representation and not a means of construc-
tion of a world.
These objectivists principles are not suitable for studying problems in
social settings, and can lead to some harmful and even dangerous conse-
quences if they are adopted in information analysis and systems develop-
ment.
3.1.2 Subjectivist paradigm
In daily business practice, and legal and other practical aairs, consensus
continually breaks down as individuals see the world dierently from each
other. This gave rise to the idea of putting the focus on subjectivism in order
to investigate whether this doctrine provides a more proper basis for study-
ing the type of problems in social domains. Whereas objectivism assumes a
single reality and explains dierences of ideas as aberrations, subjectivism
treats dierent ideas of individuals as starting points for a shared reality. In
its radical forms it turns away from pretended absolutes and origins, from
xed principles and closed systems. Subjectivism emphasises the abilities of
individuals, their freedom to choose courses of action and the moral
responsibility for their choice, as well as the uncertainty, novelty and strife
they bring about. Importance is also assigned to the aective and cognitive
behaviour of people as well as to their cognition. Reality in this stance is
experience and it is found within sensible ux. It is not ready-made but in a
making process, unnished and growing in all places where individuals
pursue their ends. The subjectivist paradigm rejects the appropriateness of
natural science methods for studying the social world and seeks to under-
stand the basis of human life by delving into the depth of the subjective
experience of individuals. The social world is studied by pursuing an under-
standing of the way in which the member of the society creates, modies,
and interprets the world to which he belongs.
Table 3.1 (Stamper 1993) summarises and compares typical objectivist
and subjectivist positions in some key concepts related to information
systems work. The subjectivist position has increasing inuences on com-
munication and information systems studies. One of the key inuences is
that the role of language is seen as not only descriptive, but also construc-
tive. The using of language is the performing of actions. A social world can
be created, altered, twisted, and moved by the use of language. For
example, performative utterances such as I do [sc. take this woman to my
lawful wedded wife], as uttered in the course of the marriage ceremony, and
I name this ship the Queen Elizabeth can create new states of aairs even
without any physical actions involved (Austin 1980). All speech acts (Searle
1969) in proper circumstances will have the eect of constructing or alter-
ing the world, for example, by creating an obligation or changing a legal
relationship in business aairs.
Table 3.1. Objectivist and subjectivist views of some key concepts.
Concept Objectivist view Subjectivist view
Reality objectively given, the same for created subjectively and socially
everyone and composed of with subtle dierences between
entities, their properties and groups of the knowing agents
relationships
Data a means of representing the a means of indicating intentions
truth about reality and coordinating actions
Truth the correct correspondence a consensus reached (temporarily)
between some real entities as a basis for coordinated action
Meaning a relationship between a sign a relationship between a sign and
and some real entity some pattern of action established
as a norm within a group
Information a kind of plumbing system a semiological system, mainly
system through which data ow informal but supplemented by
formalised messages
Role of the to specify the truth data to assist the users to articulate
analyst structure and functions of their problems, discover their
the system needed by the information requirements and
users evolve a systemic solution
3.1.3 Radical subjectivist paradigm
The research programme MEASUR adopts a radical subjectivist paradigm
(Stamper et al. 1988). In this paradigm, the reality is seen as a social con-
struct through behaviour of the agents. People in this world share certain
patterns of behaviour which are governed by a system of norms. As people
argue, communicate, negotiate and interact with each other, the world is
constantly changing. The changing world will, in turn, bring about new
values, habits and beliefs as norms. When one is trying to understand con-
cepts such as knowledge, truth, language and meaning, then agents and
actions should be taken as the starting point. This position can be formu-
lated as the following two axioms on which the major part of the research
work is based:
(1) there is no knowledge without a knower; and
(2) there is no knowing without action.
From these axioms, the well-formed formulae for knowledge representa-
tion have been built up: there will always be two components in a unit of
knowledge representation, the agent (a person or a group) and the behav-
iour:
or knower-termbehaviour-term
or
agent-term action-term.
These well-formed formulae require every piece of knowledge to be expli-
citly tied to the agent who knows it or has constructed it, and is responsible
for it. Knowing something or being able to do something becomes an item
of knowledge only from the viewpoint of a certain agent. The notion of the
universal truth, therefore, becomes irrelevant in this case. Instead, the
concept of responsibility becomes as vital as the truth in a classical logic.
3.2 The semiotic framework
Traditionally the divisions of semiotics have been syntactics, semantics
and pragmatics, which deal, respectively with the structures, meanings and
usage of signs. Stamper (1973) has added another three. Physics, as a sep-
arate branch concerned with the physical aspects of signs at the level of
signals and marks, has been further added to the taxonomy of semiotics.
It gives a handle to deal with the factors governing the economics of signs,
which has become important in business contexts. Empirics has been
introduced as another branch to study the statistical properties of signs
when dierent physical media and devices are used. At the other end of
the spectrum, he adds another layer of the social world where the eects of
the use of signs in human aairs are studied. All these branches of semi-
otics are organised into a semiotic framework (Figure 3.2).
Issues at the three upper layers in the semiotic framework are concerned
with the use of signs, how signs function in communicating meanings and
intentions, and what the social consequences are of the use of signs. Studies
of issues at the lower three layers will answer questions as to how signs are
structured and used in language, how signs are organised and transmitted,
what physical properties signs have, and so on.
3.2.1 Physics
A sign in a physical form is a phenomenon. It can be a sign in motion, which
is normally termed a signal; and it can be static, then it is a mark. The
physical properties of a sign can be its shape, size, contrast, intensity,
moving speed, acceleration, loudness, source, destination, etc., depending
on the type of the sign. Such physical properties can be studied with
methods of physics and engineering. Other related questions can also be
investigated, such as material of sign carrier, energy carried or consumed by
the signal. The direct relevance to communication and information systems
of knowing the physical properties is that people can design physical
Human
information
functions
The IT
platform
PHYSICAL WORLD: signals, traces, physical distinctions,
hardware, component density, speed, economics,
SOCIAL WORLD: beliefs, expectations, functions,
commitments, contracts, law, culture,
PRAGMATICS: intentions, communications,
conversations, negotiations,
SEMANTICS: meanings, propositions,
validity, truth, signification, denotations,
SYNTACTICS: formal structure, language, logic,
data, records deduction, software, files,
EMPIRICS: pattern, variety, noise, entropy,
channel capacity, redundancy, efficiency, codes,
Figure 3.2. The semiotic framework.
devices for storage, transmission and representation. From the physical
point of view, in databases, there are collections of physical tokens which
can be stored, moved around, for input, output and display. For an infor-
mation system or a communication network to function properly at this
level, one has to get the right RAM or disk storage, right cable connections
between personal computers and stations.
The word information has many dierent common-sense usages. In the
physical area it simply means a collection of tokens as when we talk of the
information in a database amounting to x megabytes, for example. We
might, in a practical way, talk of a signal, such as a ash of light, meaning
that there is a mark such as a transparent area on a lm. Models of infor-
mation systems at the physical level describe the available range of physical
tokens (signals or marks) available as input to or output from various phys-
ical components and they model the cause-and-eect relationships between
them, their energy and material requirements. For example, when develop-
ing information systems we model the locations and telecom links between
sites, the volumes of information in characters per unit of time which they
generate and the volumes they store, in order to discover the capacity of
channels and storage devices needed.
3.2.2 Empirics
As a branch of semiotics, empirics studies statistical properties of signs, in
which the object of study is a collection of signals or marks. Questions for
study at this level are eect of coding, entropy measurement, optimal signal
transmission, channel capacity, etc. Information from the angle of empirics
can be viewed as a stream of signals which must be transported from one
location to another, regardless of what they mean. For example, if binary
coding is used for communication between sites, we could have the follow-
ing coding:
a00001
b00011
c00101
d00110
In such communications, coding is done at the sending end and decoding at
the receiving end. Coding and decoding procedures aim at achieving
optimal transmission of information through noisy channels with high
accuracy at a rate close to the channels maximum capacity. There will
always be some necessary measures associated with a communication
process: imposing a structure on a block of messages before transmission;
transmitting some additional messages to describe the structure, and then
processing the structure at the receiving end. To get more out of the
channel, more elaborate checks, calling for the transmission and storage of
larger blocks of signals, subjected to more complex processing, are all nec-
essary. Information should be coded with the minimum number of carrier
signals for coding and transmission eciency. The word meaning in this
context means the equivalence of codes.
3.2.3 Syntactics
When there is a satisfactory range of physical tokens that provide a reliable
way of encoding diversity of use, physical representation of signs and their
statistical properties are covered by physics and empirics. We can then focus
on the complex structures of a language regardless of the actual expression
in specic terminal tokens. We can dene a syntax using names of syntactic
categories and production rules, but we are indierent about the rules rep-
resenting the terminal characters.
Syntactics is concerned with rules of composing complex signs from
simple ones. Information can be coded following a certain structure; a
complex sign, a word, a mathematical expression, or a sentence can be com-
posed of some more basic parts according to the rules. For a formal lan-
guage, terminal symbols (or vocabulary) are fundamentals of sign
construction as far as syntactics is concerned. The rules for generating and
parsing formal expressions allow us to measure the complexity of a single
formula and a sign system. When a formula can be produced or parsed in
two or more ways we have an instance of syntactical ambiguity, another
semiological phenomenon in the syntactic domain. A formula or an expres-
sion carries a piece of information. The meaning of the information at this
syntactical level can be studied using the transformation rules. Two sets of
formulae have the same meaning (paraphrase each other) if each can be
deduced from the other. The meaning of one includes the meaning of the
other if it allows the other to be deduced from it.
The rules for combining these parts are called grammar. Grammars of a
natural language teach people how to construct grammatically correct sen-
tences, but do not tell how to express semantic meanings. However, if one
wants to express oneself precisely, to follow the grammar is one of the con-
ditions.
3.2.4 Semantics
Meaning or semantics of a sign is normally considered as a relationship
between a sign and what it refers to: a sign denotes a denotatum. This is its
referential meaning. Under this denition of meaning, there has to be a
reality assumed, so that signs can be mapped onto objects in the reality.
For example, a sign table represents a kind of object with certain proper-
ties, such as having legs and a at surface. This way of treating meaning is
reasonable if the problems are simple ones where the boundaries of signs
and reality are clear and a shared consensus for example, properties of
objects is established. Meaning is then a logic function mapping words to
reality.
But in most social aairs, this is not the case: there is no agreed unique
reality; a consensus does not stay there for ever. People strive for
clarications and consensus. Once a consensus is established, it is soon
subject to question, criticism, or modication. Use of signs and languages
can clarify ideas, and can also alter or construct the world. Meaning in this
respect is more appropriately seen as a relationship between a sign and
behaviour. A sign has its behaviour, sign behaviour, behaviour in which
signs occur (Morris 1946). The meaning of a sign relates to the response the
sign elicits in a given social setting. It signies any and all phases of sign-
process (the status of being a sign, the interpretant, the fact of denoting the
signicatum), and frequently suggests mental and valuational processes as
well (Morris 1946). This is what Stamper calls the behavioural meaning
(Stamper 1973). Meaning in this sense is a result of the use of signs, and is
constructed, constantly tested and repaired through use of signs.
At the semantic level of use of language, meaning acts as the operational
link between signs and practical aairs. People use signs or a language in
communication. To enable one person to understand another, there must
be some norms governing the use of signs which are established and shared
in a language community. These rules regulate peoples behaviour in using
signs, producing sentences. Norm-conrming uses of signs are acceptable
to and understood by the members of the same language community, so
that the expected eects can be produced on the audience. To produce
eects on the audience, and furthermore to change states of practical
aairs, is the ultimate goal of using signs, but the intentional and social
eects are embodied in a sentence. In a sense, they are implied in the sen-
tence, and are an extensional aspect of such a sentence. What is rstly
reected in the sentence is the meaning at semantic level which is accommo-
dated by propositions. The sentence can be examined for its validity,
signication and correspondence to the business world. However, meaning
can only be understood if the context where the sentence is uttered is taken
into account. A word, as well as a sentence, can mean anything. The literal
meaning of a sentence, as described by Searle (1979, p. 117), is the meaning
that is independent of any context whatsoever. One can well imagine how
dicult it would be to establish such literal meanings. Winograd and Flores
(1987) show with the following example that it is impossible to establish a
context-independent basis for circumscribing the literal use of a term even
as seemingly simple as water:
A: Is there any water in the refrigerator?
B: Yes.
A: Where? I dont see it.
B: In the cells of the eggplant.
Andersens (1990) example of conversations in a car repair shop illustrates
that background information is often necessary to understand the meaning
of words, phrases and fragmented sentences. Meaning is highly dependent
on the process of using them, dependent on the behaviour of the agents.
A signs functions at three upper layers of the semiotic framework,
semantics, pragmatics and the social level, are concerned with interaction
with humans. In a broad sense, they all directly relate to the use and eect of
signs. Talking about functions, when people employ a sign or utter a sen-
tence, the rst object is to express a meaning. The next goal is to convey a
certain intention. The ultimate goal is to produce eects at the social level,
such as to set up an obligation, or alter a state of aairs. The fullment of
the objective at each former level is the basis for the latter. For example,
someone tells another person: you are stepping on my toes. The meaning
of this sentence is not dicult to understand; the intention is to inform the
addressee that he has done something wrong. But the eect of the sentence
does not stop there, the speaker is trying to change the situation of being
stepped on, and perhaps he is pushing the addressee to feel obliged to apol-
ogise to him.
3.2.5 Pragmatics
When a sign has a meaning, it can be used intentionally for certain pur-
poses, such as communication. Pragmatics, in such a case of the purposeful
use of signs, is a branch of semiotics concerned with the relationships
between signs and the behaviour of agents. Within a social community,
there exist common knowledge and shared assumptions. These basic
assumptions serve as a minimum basis for communication. However,
dierences in personal experience, values and expectation can create
diculties in understanding. As soon as there is a problem in understand-
ing, people have to come back to the minimum basis, and from there they
can expand their common knowledge and reach their new agreement. Dik
(1989) calls the personal possession of knowledge and experience prag-
matic information.
A process of verbal communication between two people can be modelled
like this. The speaker and hearer each have their own pragmatic informa-
tion base. The speakers pragmatic information base allows him to form an
utterance with an expectation that the hearer will understand both seman-
tic meaning and his intention. For example, by saying Im hungry, the
speaker expects the hearer will then interpret the sentence, with his prag-
matic information, understanding that the speaker is actually saying Its
time for lunch, lets go to the cafeteria. If two participants in a conversation
are from the same cultural community, they have a large amount of shared
pragmatic information. In this case, they can each assume the pragmatic
information of the other. Knowing pragmatic information of the opposite
side can shorten the route of conversation and improve the eciency. On
the other hand, incorrect assumption of the pragmatic information of the
opposite side can lead the conversation astray. Therefore, it is important for
a speaker to know what assumptions he shares with the hearer before com-
munication. Dik (1989) suggests that the speaker should start from shared
information, and proceed from there to estimated non-shared information
in order to have this added to, or substituted for, pieces of the hearers infor-
mation. The shared or non-shared pragmatic information is mainly deter-
mined by the cultural dierences.
Apart from the shared knowledge, the context where the communication
takes place is important to the pragmatic eects. The following elements
can be considered in the context: speaker, hearer, intention, purpose, theme,
time, location, and psychological states of the speaker and hearer, such as
intentions, desires, beliefs, etc. Relative positions of the speaker and hearer,
as discussed by Searle & Vanderveken (1985), are also an important aspect
of the context. For example, suppose there are two dierent relative posi-
tions of the speaker and hearer: in one, the speaker has an authority over
the hearer and, in the other case, the two are peers. An utterance of the sen-
tence Leave the room produces dierent pragmatic eects. In the former
case, it may be an order while in the latter it may be a request. The speaker
has an intention when he begins the conversation by uttering his speech
act. The following words, for example, either in verb or other forms, can
characterise intentions: report, inform, announce, declare, suggest,
propose, require, order, allow, permit, inquire, suppose, assume,
hypothesise, apologise, etc. When the speaker in his utterances uses such
words, one can know the speakers intention as whether he wants either to
report his observation or to require something. But in an actual conversa-
tion, these words may not need to be used and the speaker is still able to
express his intention. In such cases, analysing the intention by means of
formal modelling is dicult. However, it is always possible for the hearer to
grasp the speakers intention by applying his pragmatic information with
intuition in an informal way. Some techniques for capturing intentions will
be found in Chapter 9, where the semiotic method for systems analysis is
presented.
The purpose of the conversation is set up by the speaker and is related to
the speakers intention. The speaker can conduct a series of speech acts to
pursue the purpose which may be clearly stated by him or hidden. For
example, if the speaker suggests something to the hearer, his purpose in
carrying on the conversation may not be simply to suggest but to persuade
the hearer to join him in some action. Such a purpose is dicult to detect
from the beginning of the conversation; it may become clear to the hearer
as the conversation develops further. The hearers perception of the
purpose can change in the course of the conversation. In this example, at
the beginning of the conversation: the hearer may understand it as merely
suggest. But as soon as he changes his perception of the purpose of the
speaker, he may interpret suggest as persuade. The purposes of commu-
nications are dicult to model by formal means. However, we can under-
stand the underlying mechanism by studying social and cultural norms at
the social level. The theory of functional grammar (Dik 1979, 1989) pro-
vides an elaborate method of handling pragmatic functions of sentence
constituents such as theme, tail, topicality and focality. The analysis of
these constituents is useful for one to establish a view on functions of a sen-
tence in relation to the wider communicative setting in which the sentence is
uttered. It can help one to obtain a continuous picture of the ow of com-
munication. Time and location are also important aspects in studying com-
munication eects. Eects of the communication acts cannot be studied in
isolation, they have to be related to surrounding communication phenom-
ena, for which time and location are aspects of the context indispensable for
understanding.
3.2.6 The social level
When a conversation takes place between two or more people, a change at
social level will be caused. A conversation can be seen as a proper chain of
speech acts. It begins with the speaker expressing some intention in an illo-
cutionary act. As soon as the speech act is addressed to the addressee, an
obligation is usually already built up for the addressee that he has to
respond regardless of the content of the conversation. For example,
suppose a person simply says Good morning to a small group of col-
leagues. Following social convention, there is an obligation immediately on
individuals of the group to acknowledge his greeting. If someone did not
do so, he might not feel at ease. In a social setting, norms or social conven-
tions govern peoples behaviour. Communication must follow a certain
pattern. Any deviation from the normative patterns may be unexpected,
and may either require an excuse or otherwise be regarded as unacceptable.
For example, an invitation can lead to either acknowledge but not accept,
or accept. The obligations are established on both sides by the speaker and
the addressee. As soon as the invitation is issued, the speaker must be pre-
pared to be able to entertain the guest. In case of acceptance, the invitee has
actually made a promise to the inviting person that he will full his obliga-
tion, i.e. going to the party or dinner at the agreed time. Meanwhile, the
inviting person has committed himself to receiving the guest. The interpre-
tation of the communication acts, under a given circumstance, produces
social consequences.
The process of performing communication acts is sometimes a complex
process of invoking, violating, and altering social norms. Consciously or
not, people interact with each other and make interpretations according to
norms, while sometimes they deviate from the normative for their own
interest. As Stamper (1973) remarks, for man, a social animal, the norms
are a biological necessity. Norms have long been classied into several cate-
gories: perceptual, cognitive, evaluative and behavioural. Perceptual norms
are implicitly agreed ways of seeing the world. With these norms, words can
be assigned meanings that enable people to use them as labels on the hap-
penings around them. People can code the message picked up by the eyes
and ears in a way that is useful to other people. Cognitive norms are the
standardised beliefs and knowledge possessed by a group. These norms
ensure that the members of the group acquire the knowledge and expecta-
tions about the world which have been accumulated by other members of
the cultural group. Evaluative norms direct the group towards common
ends. They provide a framework of valuation with which peoples behav-
iour can be assessed. Behaviour norms govern people so that they behave in
an appropriate manner in a given cultural setting.
The nature of a group to which people belong can vary from just a cul-
tural group, an informal meeting group, a gang formed in the street, or a
business establishment, to a government organisation. Accordingly,
various degrees of formality of norms can be found. For a casual, informal
group, very often the norms are informal while in the case of a highly for-
malised group, such as an administrative department of government or
business organisation, most of the norms are formal. Besides, there is a
horizon of norms being explicit or implicit. Table 3.2 shows examples of
these types of norms.
3.3 An example of semiotic analysis
At a strategic level, semiotics can guide one in understanding how an
organisation works as an information system. Operationally, it can also
help one in analysing how to make a simple communication process such as
a telephone conversation successful. Here a simplistic example of a tele-
phone conversation is used to illustrate how a simple Semantic Analysis can
be conducted to diagnose possible problems in the communication.
A successful telephone communication is determined by the factors in six
semiotic aspects.
At the physical level, the telephones must be connected by the phone line
through telephone service providers.
At the empirical level, the voice signals will be converted into electronic
(or optical) signals and transmitted between two telephones.
These two levels are the technical infrastructure which is provided by the
telephone companies. They are normally not the concern of the users.
At the syntactical level, the two people involved in the telephone conver-
sation must follow the same grammatical rules, i.e. speak the same lan-
guage.
3.3 An example of semiotic analysis 35
Table 3.2. Types of norms.
Formal Informal
Explicit ocially documented norms, unocial but agreed verbally, e.g.
e.g. laws, legislation, each one sharing this big oce will
organisational regulations buy coee supply in turn
Implicit not applicable habit or convention never verbally
mentioned but people follow, e.g. the
Club members wear formal dress
when attending their monthly meetings
At the semantic level, the words, the technical and non-technical terms,
and the things referred to in the conversations must be understood by
the two people. The sentences and the contents of the conversation
must make sense to both of them.
At the pragmatic level, there is a concern with the intention (of the caller),
and there may be silent messages beneath the surface. For example, if
person A calls person B and says Im interested in your goods but the
price is a bit too high, As intention will be to ask whether B can lower
the price a bit.
At the social level, social commitments and obligations can often be
created or discharged as the result of a conversation. Following the
above example, if B answers Youll get ten per cent discount if you buy
ten or more of the PCs, there will be an obligation on B to give the dis-
count if A buys ten or more PCs.
A communication, from the callers point of view, can be said to be success-
ful only if his or her messages are understood by the listener; intentions are
learned by the listener; and social purposes are met.
4
A semiotic approach to information systems
development
Development of a computerised system normally goes through several
stages, and this whole process is termed a system development lifecycle.
There are various lifecycle denitions (see Leslie (1986), Macro & Buxton
(1987), Olle et al. (1991) for example). Many researchers have been engaged
in bettering methods for each of the phases in a lifecycle. Some of the
methods are aimed at one of the phases while others are claimed to be
useful for more than one of them. Recently, a great deal of attention has
been paid to user requirement analysis and specication. This is because, on
examining the sources of diculty for large systems projects, we nd that
the incorrect code is a relatively insignicant factor, dwarfed by incorrect
functional specication and (even more signicantly) incorrect require-
ments (Boehm 1981; Goguen 1992).
There are many ways to phase activities in systems development. Macro
& Buxton (1987) point out that many models may be excessively simple
while others are extremely elaborate. In general, activities of systems devel-
opment can be grouped under a few headings: conceptualisation of systems
scope, requirement analysis, systems design, implementation, validation,
acceptance, maintenance, and, nally, obsolescence (which may be fol-
lowed by activities of developing a new version of the system). The research
programme MEASUR (Stamper 1993) has developed a set of methods to
deal with all aspects of information systems development.
4.1 MEASUR
MEASUR, a radically new set of norm-oriented methods for business
systems modelling and requirements specication for software develop-
ment, is the result of a research team working in the eld of information
systems since the later 1970s. The concept underlying MEASUR is that
37
organisations, themselves, are information systems and the social norm is
the appropriate unit of specication. MEASUR provides ve major
methods to cope with the tasks in these phases.
Problem Articulation Methods (PAM) These comprise a set of methods
that can be applied at the initial stage of a project when one encounters
a vague, complex problem. The methods help the user to identify issues
worthy of attention. In the case of solving a problemor a project for an
innovation, the central task will be identied as the focal system. The
situation in which the problem resides and the envisaged system must
operate is seen as the infrastructure, and will be treated as a composi-
tion of many collateral systems. In order to characterise the problem
situation, the user of the method will be assisted in dening unit
systems whichhave a start, a nishandaordances whichare valuedby
the stakeholders (the user himself may be one of the stakeholders). By
using the method, undesirable omissions from analysis and specica-
tion can be reduced. The results can provide a basis for socio-economic
analysis, and lifecycle and project control. There are four specic
methods under the heading of ProblemArticulation(Kolkman1995):
(1) Unit System Denition, a method which provides handholds to
describe an action course and list interdependent agents who
might take an interest in it;
(2) Valuation Framing, a method which can be used to reveal the cul-
tural behaviours of the constituents in relation to the benets and
drawbacks of an action course;
(3) Collateral Analysis, a method which assists in structuring a
problem situation into a kernel course of action and surrounding
or collateral activities and subsequently gives names to each part;
(4) System Morphology, a method which can be used to clarify three
basic functional areas of social systems. Each of these compo-
nents can in turn be treated as a unit for continued analysis.
Semantic Analysis Method (SAM) This method takes an articulated unit
system or focal problem as the input of analysis. Within the agreed
boundary of the focal problem, one of the major objectives of
Semantic Analysis is to assist the user or problem-owner in eliciting
and representing their requirements in a formal and precise form.
Possibly with the help of a facilitator, required functions of the envis-
aged system will be specied in the ontology model which describes an
acceptable view of the responsible agents in the focal business domain.
The meaning of a word used in the semantic model to represent the
38 A semiotic approach to information systems development
business world is treated as a relationship between the word and
appropriate actions.
Norm Analysis Method (NAM) This method gives a means to specify the
general patterns of behaviour of the agents in the business system. The
analysis of the regularities of behaviour is focused on the social, cul-
tural and organisational norms that govern the agents actions in the
business domain. A norm can dene responsibility for an agent occu-
pying a certain incumbency (e.g. a person playing a certain role); a
norm can specify conditions in which some action may (or must, must
not, etc.) be performed by some agent. The norms can be specied in
NORMA, a knowledge representation language, and further be trans-
lated into a computable language, LEGOL, for further processing.
The systems specication can be directly implemented in the dedicated
software development environment, Normbase, or be implemented in
any other programming languages. Each specied norm is associated
with a pattern of actions described in the computer system. For a com-
puter information system, the norms are eectively used as function
denitions or constraints on system functions.
There are two more methods: Communication and Control Analysis, and
Meta-Systems Analysis. The former assists one in analysing the various
communications between all the responsible agents and unit systems
(identied with the Problem Articulation Method) and within the focal
system. The communication messages are classied as informative, coordi-
native, and control, according to the intentions of the sender. The norms
will be added to the governed procedures, message passing, rewards and
punishments. The latter of these two methods, Meta-Systems Analysis,
treats the whole project of innovation or system development as the object
of study. It requires one to step outside the object system and helps one to
treat the meta-problem in planning and management of the project.
4.2 How MEASUR can help in information systems development
There is no absolutely xed sequence of the activities engaged in an infor-
mation systems project; and therefore the application of MEASUR
methods need not follow a xed order. However, the following major
phases are often found in a systems development project: infrastructure
analysis, systems analysis and systems construction; MEASUR methods
can be used in each of the phases. This section will describe how an infra-
structure analysis can be conducted, with illustrations of a number of
4.2 How MEASUR can help 39
MEASUR methods. The discussion of systems analysis, design and imple-
mentation will be brief in this chapter, because they will be the main focus
of discussion in the following chapters.
4.2.1 Infrastructure analysis
The objectives of this phase are to understand the social, organisational,
cultural and technical contexts, to understand business objectives and
strategies in the organisation, and to identify business and technical prob-
lems and the feasibility of solving the problems. To achieve these objectives,
one can apply the semiotic framework for a semiotic diagnosis, followed by
the Problem Articulation Methods (PAM).
Semiotic diagnosis
A semiotic diagnosis examines the organisation as a social system which is
constructed through the use of information. The main steps of diagnosis
are the same as the layers described in the semiotic framework (Figure 3.2).
The framework links the physical and social worlds through the informa-
tion mechanisms people use in the organisation, rather than separating the
issues into problems for the experts on production engineering and logis-
tics, versus problems for the experts on social psychology, industrial rela-
tions and organisations. Let us see how each of the levels in the semiotic
framework can be used.
Physical problems concern the media for carrying and storing signals.
Word of mouth, paper, electronic devices are examples. The diagnosis
at this level will focus, for example, on whether stocks contain the same
information as in the inventory, placement of machines matches the
plan, and production processes are the same as reported. As far as
the informal information system is concerned the physical locations of
the participants are crucial to their communication. Thus architecture
and travel arrangements are important aspects of the physical aspects
of communication. Business issues arise because of the economics of
the physical resources. Information technology can often enable us to
substitute a cheap medium for an expensive one (email for snail-mail)
but it may even allow us to eliminate several lorries and a depot full of
stock by using information more skilfully.
Empirics is concerned with the problem of the unexpected, the random-
ness of the entropy which so-called information theory treats as the
root of all information. When you run an organisation you have to
cope with many dierent streams of events that are loaded with ran-
domness. Therefore we organise to turn unpredictable raw materials
into uniform products, to give clients a predictable service to cope with
their unpredictable problems. You try to keep the plant running
smoothly despite randomness in the supply of labour, the functioning
of machines, failures of the transport systems and the irregularities of
demand. You also want to generate diversity and unpredictability in
order to innovate, keep competitors guessing and keep on your toes.
Business issues are frequently prompted by considering the manage-
ment of uncertainty overcoming it or generating it. These are not
necessarily primarily concerned with information systems; but one
should not forget that the ultimate goal of having information systems
is to help with the management uncertainty.
Syntactics is the aspect of semiotics concerned with structure. At one
level it concerns the structure of sentences in a language but it can be
treated much more broadly for this diagnostic purpose. The activities
we have to manage are often chaotic and the issues of interest here are
how to impose order and regularity, and hence operate far more
eciently. That is what production engineering is about in the factory
and also systems analysis in its oce analogue. One looks for the right
kind and degree of formality. Only when one has succeeded in this can
one introduce a computer system to support.
These three levels are all concerned with the eciency issues. This class of
problems can normally be solved using technology. They receive nearly all
the attention when classical methods of requirements analysis are used.
MEASUR is able not only to draw attention to the next three levels but also
to provide methods for tackling the problems they raise. They are always
information oriented problems, and they are becoming continually more
important.
Semantics concerns meanings. When people from dierent backgrounds
have to work together they need to understand one another. On a
smaller canvas we can depict semantic problems that organisations
constantly encounter the need to respond to a new technology, a new
marketing concept, a social change aecting the consumers, a new
organisational concept. Eectiveness is a matter of doing the right
thing and that entails a continual search for such new ideas. They all
throw up problems of semantics.
Pragmatics concerns intentional behaviour. Signs are used to indicate
intentions. Doing business is largely a mater of getting things done
through signs that are both meaningful and charged with intention.
The characteristic of an organisation with pragmatic problems is
muddle. Sometimes this is caused because there is no consensus about
goals, the commonly accepted intentions of a team, but sometimes the
muddle is caused by misunderstandings about intentions. Issues in this
area have to be resolved at the highest level in strategic planning but
they also have to be resolved at the micro-level in the design of the
organisation and its information systems.
Social problems are essentially those about how people do or should
behave, that is problems of norms and expectations. The culture of the
organisation is signicant, because all information systems are ulti-
mately dependent for any of their capacity to use signs upon the
culture in which they are embedded. MEASUR is based on the analy-
sis of the norms of an organisation, so MEASUR takes the social level
as the starting point and is ideally suited for elucidating issues in this
area.
MEASUR concerns itself with the three upper levels in the semiotic frame-
work, those constituting the human functions. Thus, the use of methods is
not constrained by any technical platform on which the human information
system stands.
Organisation and context valuation
The next major task in the systems development project may be to conduct
a total valuation of the organisation and the context for the proposed
project. This analysis can take place later if the impact of the project is not
considered a crucial and urgent concern at this stage. This valuation can
also be conducted in parallel with other analysis.
All the stakeholders involved will be identied and the valuation will be
done taking into account the interest of all stakeholders. The total system
will be the object of valuation. Let us imagine that a company is going to
initiate a project to develop a computer information system. The stake-
holder groups in this case will include the owners of the company, the man-
agement, the operational sta, the IT department, the clients, suppliers,
competitors, bystanders, the general public and so on. Each group will have
a cultural system that governs how it will value this kind of innovation. The
dierent stakeholders may react dierently to the proposed project. To
enable a comprehensive valuation, valuation framing has been developed
which is based on an anthropological classication of cultural norms
(Figure 4.1). The management, for example the stakeholder C in the gure,
may see the proposed IT project, in the aspect of exploitation, as an oppor-
tunity for improving the productivity, which is viewed dierently by the
operational sta.
The valuation framing can also be applied at a more rened level such as
the unit systems that comprise the total system.
Relationships between unit systems
The method of collateral analysis helps in analysis and articulation of rela-
tionships between unit systems that comprise the large, complex system.
PAM takes the central task in an innovation, such as an IT project, as the
focal system. The scope of the analysis is the focal system and its infrastruc-
ture in which the focal system resides and operates. Suppose the proposed
IT system has been approved to go ahead, then the focal system will be the
new IT system in operation. But it does not spring out of nothing and it can
only function successfully in the right infrastructure. Collateral analysis
forces one to take the total system apart into a number of unit systems
around the focal system. The collateral systems are interrelated and form
an infrastructure for the focal system (Figure 4.2).
S
t
a
k
e
h
o
l
d
e
r

A
S
t
a
k
e
h
o
l
d
e
r

B
S
t
a
k
e
h
o
l
d
e
r

C
S
t
a
k
e
h
o
l
d
e
r

D
Subsistence
Classification
Territoriality
Temporality
Leaning
Play
Protection
Exploitation
Association
Interaction
Stakeholder Cs
valuation of a
project in the aspect
of exploitation
Figure 4.1. Valuation framing for organisation and context analysis.
The collateral analysis method takes the fuzzy, total system and locates
its eective boundary in a given environment. The method, as a checklist
for any innovation, guides people to identify unit systems and organise
them into the collateral structure. This method will be most valuable if the
social and technological infrastructure is poor. An innovation project may
become unexpectedly complex and a larger scale operation than expected.
Just consider placing a new plant for electronic products in the silicon
valley of any of the economically advanced nations compared with
launching it in a developing country. The lack of a piece of electric wire, for
example, may make it impossible to connect a personal computer to a
power supply, and therefore will make the machine fail to work in a remote
undeveloped area. In order to bring the focal system into operation and
sustain the situation, the analysis for planning and management of the
innovation project must cover all these collateral systems.
System component analysis
For this purpose, the method of systems morphology can be applied as a
complement to collateral analysis. Systems morphology looks inward at the
Environment
Description
Input
Output
Predecessor Focal system Successor
Fallback Recovery
Launching
Backup
Terminating
Constructing
Available
Dismantling
Resources
Figure 4.2. The collateral analysis method.
structure of each unit system. The approach taken is similar to the top-
down, structured methods for requirements analysis, but the focus of
attention is radically dierent. It focuses on the identication of norms that
govern peoples behaviour in a unit system. Each unit system can be divided
into three subsystems: substantive, communication and control subsystems
(Figure 4.3). The substantive behaviour is governed by the assignments and
tasks, rules and norms, that are derived from the organisational objectives
within a given institutional structure. Results of the actions in this category
are supposed to contribute directly to the attainment of the business goals
of the organisation. The actions will normally result in changes physically
or socially.
Communication (message passing) is the second type of behaviour, which
is concerned with signs, the input and output of the communication
process. Communication norms direct the passing of messages from one
agent to another for the purpose of supporting successful performance of
the rst type of actions. Signs are employed and sent between agents with
agreed meanings by both parties involved in communication to express
intentions. People who have a substantive task to perform will communi-
cate anyway. The norms in this category are those which supplement this
informal kind of communication by adding the formal exchange of mes-
sages. When people are widely separated by space and time, the informal
system is less eective. This is obviously necessary between organisation for
most inquiries, orders, invoices, reminders, etc. However, it is quite easy to
introduce more formal communication than necessary for running the busi-
ness eectively. One way of re-engineering is to look at the communica-
tions, possibly with a view to making better use of the informal system.
The third type of organisational behaviour, control, is governed by some
enforcement or inducement norms. Within an organisation, the power of
unit system x
substantive message
about message x.m.s
substantive x.s
x.s.s
message-passing x.m
control x.c
x.s.m x.s.c
messages about
messages x.m.m
control of
messages x.m.c
Figure 4.3. Systems morphology for analysing unit systems.
enforcement may be stipulated in rules and regulations. Between organisa-
tions, it may be generated from inter-rm agreements, treaties, or contracts.
But the power of enforcement both within and between organisations ulti-
mately rests upon socially established norms which may be underpinned by
the law. This control part of behaviour within a rm ensures every relevant
agent acts properly in performing the substantive tasks. If an agent under-
stood what his duty is but failed to comply with the rule, the organisational
control will be imposed on the agent. Sanctions (punishment and reward)
are necessary means to maintain social and organisational orders as well as
incentives for achieving the goals. Once again we should be prepared to
adjust the balance between formal and informal control to increase
eectiveness.
4.2.2 Systems analysis, design and implementation
The result of the infrastructure analysis gives us the power to explain the
situation and articulate the innovation project into a better dened and
structured problem. It also generates sucient terminology to represent all
the relevant actions. The major tasks in the further stages of a systems
development project will be as follows.
Requirement analysis, which includes information analysis and func-
tional analysis;
Construction, which includes systems design and implementation;
Operationalisation, which includes the use, monitoring and maintenance
of the system.
This section briey describes how MEASUR can help in coping with these
tasks; details will be found in the following chapters.
The Semantic Analysis Method (SAM) can be used to elicit and specify
users requirements. The method requires one to start by identifying the
agents in the problem domain. The agents are people in dierent roles, with
dierent responsibilities and authorities in an organisation. A semantic
model, as the result of analysis, will capture the possible actions (or pat-
terns of behaviour) of the agents. These patterns of behaviour are called
aordances, the actions that are aorded by the agents, and that the agents
are potentially capable of doing. The aordances depend on the agents,
which is called ontological dependency. Because the information about
which agent is capable of doing what action is the most important content
in the semantic model, it is also called the ontology model.
The Norm Analysis Method (NAM) can then be used to capture norms
(i.e. rules, regulations, etc.). The norms act as conditions and constraints;
they govern agents behaviour, normally in a prescriptive manner to decide
when certain actions are performed. Norms, in conjunction with the
semantic model, dene clearly the roles, functions, responsibilities and
authorities of agents
The knowledge representation language used for semantic modelling and
norm specication is called NORMA (a language for describing NORMs
and Aordances). The specication can be translated into a computation
language LEGOLif a computer system is to be implemented.
A semantic model, as the result of Semantic Analysis, not only is rich in
semantics, but also contains information such as temporal and ontological
dependencies between agents and aordances. All these information fea-
tures can be captured in a pre-dened, uniform data structure in the
Semantic Temporal Database (STDB). The STDB constitutes the most
important foundation for an integrated software environment, Normbase,
which supports Semantic Analysis and Norm Analysis, and ultimately pro-
duces a computer information system that is based on semiotic principles
and MEASUR methods.
4.3 Summary
A project of information systems development may pass through a number
of major phases, each of which comprises some activities. These activities
and phases are not totally sequential. Many of them take place in parallel
and may re-occur a few times during the project. Table 4.1 shows major
phases and activities, and the MEASUR methods that can be used to assist
the activities.
The MEASUR methods can be used in virtually any sequence and may
4.3 Summary 47
Table 4.1. MEASUR methods and major systems development activities.
Major phases Major activities MEASUR methods
Infrastructure analysis business strategy analysis PAM
information planning PAM/SAM
Requirement analysis information analysis SAM
functional analysis NAM
Construction design Normbase
implementation Normbase
Audit audit PAM, SAM, NAM
be mixed with other techniques at many points. For example, at the design
stage, one may turn to using the entityattributerelationship method, or at
the implementation stage one may adopt a fourth-generation language or
an object-oriented language. But to use the MEASUR methods through-
out all the phases may potentially bring in great benets to the entire
project. This is because, rst of all, all the methods are consistent with one
another and based on the same underlying principles. The result of analysis
at an earlier phase always lends itself readily to the MEASUR method at
the next phase. Secondly, there are software tools developed to assist activi-
ties at all stages. The Problem Articulation Tool (PAT) is a computerised
analytical facility. With the help of PAT one will be guided to dene all the
unit systems, to identify the relationships between the collateral systems
and the stakeholders, to assess the business strategy of the organisation,
and to establish project plans. One can also use PAT for project manage-
ment if the plan is constructed with this tool. Normbase is the comput-
erised environment which provides support to cover the whole range of the
systems development activities. It supports the whole process of a Semantic
Analysis, a Norm Analysis and production of the systems documentation.
5
Knowledge representation and information
analysis
Information analysis is also called information requirements analysis, or
requirements studies. It shares with knowledge representation many funda-
mental concepts and principles. Similarities are found in techniques and
methods for both. They both address how people understand the world and
how to represent their understanding. Meaning of symbols and languages
used in modelling is a central issue for both.
All men by nature desire to know said Aristotle in his book
Metaphysics. As an agent experiences in the ux of the world in which he
acts, his faculty of sensation enables him to feel and observe aairs
around him. His memory helps him to accumulate the observations and
from them he learns about the world and knows what he can act upon in
given circumstances. Many philosophers and scientists, such as psycholo-
gists and linguists, have been preoccupied with the study of knowledge. As
computer scientists have joined in the inquiry, new perspectives of the
study have been suddenly broadened. Knowledge representation has
become one of the most important areas of study, because it is directly
related to the exploitation of human knowledge by means of articial
intelligence, (deductive) databases, and other computer-supported
methods. Each theory of knowledge (epistemology) presupposes a seman-
tic theory; otherwise knowledge cannot be captured and represented with
any language or sign system. Knowledge representation is about mean-
ings, which are relationships between signs and the other things that they
stand for. It in many respects shares the same concepts and principles as
information analysis.
This chapter will look at some considerations and approaches in
knowledge representation, followed by a discussion on information analy-
sis.
49
5.1 Some basic considerations in knowledge representation
In philosophy or epistemology, it is often thought that knowledge is indefea-
sible, justied true belief (Bradley & Swartz 1979), though there is no agree-
ment on what counts as justication, as exhibited by Lacey (1976). In
Knowledge Representation (KR), knowledge is about any kind of belief a
rational person might hold; therefore, the exercise may be more suitably
named as conceptual modelling (Way 1991). The conceptual modelling
deals with questions about the knowledge or beliefs that are held by some
agents. The beliefs are concerned with the status of universals, abstract
ideas, and concepts which are supposed to be reected in a model.
5.1.1 Expressive adequacy and notional eciency
A formal language or method for knowledge representation is devised with
the intention of encoding knowledge and describing practical aairs. The
notion of expressive adequacy questions whether a method is capable of
describing all kinds of information needed in knowledge representation. A
few crucial elements have to be taken into consideration when a formal lan-
guage or method is developed. Knowledge or belief can be subjective. This
subjectivity must be capable of being expressed in a knowledge model.
General knowledge and belief can evolve or change (for example, human
knowledge about the universe); knowledge about particular events is more
dynamic. Therefore, it is important that temporality of knowledge be
expressed in a knowledge model. In a given circumstance, knowledge can
help the agent to make decisions. The decision-making process can be a
quite complex one, in which knowledge about the past and predictions and
hypotheses about the future will be involved. Modalities (typied by the
auxiliary verbs in English such as must, could, may, shall, etc.) will be
needed in such kind of reasoning processes.
The notational eciency of a representation language concerns the syn-
tactic usage of the language. An important question regarding eciency is
how easily the language can be understood and worked with. This is often
referred to as conceptual eciency. Another consideration is the computa-
tional eciency.
5.1.2 Semantic primitives
The primitives of any system are a set of basic units or structures which
cannot be analysed further into component parts and of which all other
50 Knowledge representation and information analysis
complex structures are composed. Brachman (1979) identies ve levels of
primitives in KR: implementational, logical, epistemological, conceptual
and linguistic. Apart from the primitives of these categories, Way (1991)
suggests another type of primitives that are at the level of the presentation
scheme itself. At this level, the primitives would be how the general struc-
ture of the knowledge is conceived, for example, as predicates organised in
the logic approach, as semantic nodes and links in the semantic networks,
and as frame structures in the frames. The representation scheme of any
method, as a whole, has to consider how the primitives of the rst ve layers
are treated. Another notion of primitives in the representation scheme,
which must be distinguished from the syntactic notations, is the semantic
primitives or basic units of meaning. In her book, Way (1991) quotes Wilks
explanation of semantic primitives (see Wilks (1978)):
Primitives are philosophical atoms, usually corresponding to words in a natural lan-
guage such as English, and are said to name basic concepts underlying human
thought.
A semantic theory, which every KR method must have and be associated
with, will enable one to express ones beliefs with the primitives in the
formal notations used in the method, and also enable one to understand the
model. Meanings specify the correspondence between the signs and the ref-
erents. The semantic theory can also be a guide to eliciting knowledge from
verbal and non-verbal actions and expressions, and organising them into
the formal representation. This means that the knowledge representation
would be trying to nd meta-models for the models of the world that
underlie our everyday reasoning. Such a model is an ontological model: it
consists of a catalogue of objects and properties with law-like relations
between them (Way 1991). Way calls this approach scruy, which is typi-
cally in contrast with some axiomatic approach which has been conned
within the consequences based on axioms whose truth depends only on
guesswork.
1
5.1.3 Types of knowledge
Ryle (1949) distinguishes two types of knowledge: declarative and proce-
dural. The former knowledge is knowing that while the latter is knowing how.
5.1 Some basic considerations in knowledge representation 51
1
Sowa (1984) comments that the scrues versus the neats have dierent views of AI. The neat
view assumes that a few elegant principles underlie all the manifestations of human intelli-
gence. Discovery of those principles would provide the magic key to the working of the
mind. The scruy view is that intelligence is a kludge: the ways people solve problems are so
much dependent on ad hoc approaches and heuristics that no universal principles can be
found.
Declarative knowledge is theory-oriented. It consists of an explicit descrip-
tion of relations between concepts. The relations can be denitional depen-
dencies in terms of concepts, and functional dependencies between
quantities. This type of knowledge can be conveniently represented in many
representation methods, for example, semantic networks and frames.
Procedural knowledge is practice-oriented. It often consists of a set of pre-
scriptions for actions in relation to certain types of conditions. They are
referred to as situation-action rules or production rules. Rule-based systems
form a typical example of a suitable representation approach for this type
of knowledge. Winograd (1975) summarises the arguments in the proce-
duraldeclarative controversy as follows:
Economy Procedures specify knowledge by saying how it is used, and
every use requires a dierent procedure. A declarative approach
requires only a single copy for all uses.
Modularity Procedures bind knowledge and control in a single package.
By keeping them separate, a declarative approach makes it easier to
update and generalise the knowledge base.
Exception handling Procedures can do anything, and problems that are
not covered by the formal theory can often be handled by an ad hoc
piece of code. A declarative approach may nd diculty with excep-
tions that were not anticipated by the theory.
All the three arguments are important considerations for anyone who
develops a method for knowledge representation. Fortunately, advantages
and disadvantages of the two representation strategies are recognised by
the AI community. Many methods developed so far can also be charac-
terised as being suited to more one type than another. But since the proce-
duraldeclarative controversy arose over the role of procedural knowledge
in intelligence in the late 1970s (cf. Winograd 1975), many developers try to
cope with both types of knowledge in their methods, because of the neces-
sity for both.
5.2 Representation approaches
5.2.1 Typical examples
There are some well-recognised typical approaches to knowledge repre-
sentation, such as the logic approach, semantic networks, frames, rule-
based methods and conceptual graphs. The book of Ringland and Duce
(1988) oers an overview of those typical approaches and some research
issues.
The logic approach is mainly based on rst order logic, though other
non-classical logic, e.g. modal logic, has a role to play there. The system of
logic is known as denite Horn clauses, which are essential rules of the
form:
A if B
1
and B
2
and . . . B
n
Such a clause has exactly one conclusion A, but zero or more conditions B.
Propositions are basic elements and can be translated further into several
predicates. Propositions have truth-values that are determined by a certain
interpretation function. That is to say, any proposition must be TRUE or
FALSE under the interpretation which sets up a correspondence between
the symbols of the language and objects or values in some domain. The
predicates can be connected with connectives (e.g. (and), (or), (not),
and (implies)) and quantiers (i.e. (existential quantier), and (uni-
versal quantier)). Thus, take a piece of knowledge as follows:
John gives a book to Mary
The constituent concepts can be rst of all identied as John, Mary,
book, and give. A decent representation can be obtained as follows:
2
giving(e) e is a giving event
agent(e, John) John was the giver
recip(e, Mary) Mary was the recipient
b, book(b) object(e, b) a book is the object of giving
time(e, T) the giving was at time T
e,y,t,x giving(e) recip(e, y) if y was the recipient of x at time t
time(e, t) object(e, x) then he becomes the owner for all
(t1, t1t owns(y, x, t1)) time after this.
Semantic networks are designed as a means to describe the concepts behind
word meaning and the ways in which such meanings interact. A network is
a net or graph of nodes joined by links. The nodes in a semantic network
usually represent concepts (e.g. BOOK, GREEN) and the links (or labelled
directed arcs) usually represent relations (e.g. a book IS COLOURED
green). Semantic networks may be loosely related to predicate calculus by
following the substitution: terms and relations replace nodes and labelled
directed arcs.
Frames are ways of grouping information in terms of a record of slots
and llers. The record can be thought of as a node in a network, with a
2
Based on Pavelin in Ringland & Duce (1988).
special slot lled by the name of the object that the node stands for and the
other slots lled with the values of various common attributes associated
with such an object. This approach can be conveniently suitable for repre-
senting knowledge about structure domains or structured knowledge. The
intuition behind the frames is that the human brain is less concerned with
dening strictly the properties that entities must have in order to be consid-
ered as exemplars of some category, and more concerned with the salient
properties associated with objects that are typical of their class.
5.2.2 Conceptual graphs
Charles Peirce had many extensive treatments of graphs. He proposed that
a graph can be used as a consistent system of representation, founded upon
a simple and easily intelligible basic idea. It can represent facts, concepts
and relationships iconically (see Peirce (193158)). Graphs have recently
been used for developing many methods of knowledge representation, such
as semantic networks (see Brachman (1979), Randal (1988)), conceptual
dependency graphs (Schank 1972, 1975), knowledge graphs (Hoede 1986,
Smit 1991), and conceptual graphs (Sowa 1984). Graph methods generally
claim to capture some deeper cognitive properties than syntactic structures.
Many studies in this direction try to discover some basic concepts and types
of relations so that meanings of words can be described by case relations
(i.e. semantic relation primitives linking verbs to some other structures and
elements). For example, Schank (1972) identies a set of primitive actions,
states, and conceptual cases. Knowledge expressed in natural language
can be represented by employing these primitives.
Conceptual graphs are devised as a language of knowledge representa-
tion by Sowa (1984), based on philosophy, psychology and linguistics.
Knowledge in conceptual graph form is highly structured by modelling spe-
cialised facts that can be subjected to generalised reasoning. In the graphs,
concept nodes represent entities, attributes, states, and events, and relation
nodes how the concepts are interconnected. A concept is drawn as a box, a
conceptual relation as an oval; an arc with an arrow links a box to an oval
(see Figure 5.1).
It can also be expressed in a linear form; then the boxes may be abbrevi-
concept1 relation concept2
Figure 5.1. Concept and relation in conceptual graphs.
ated with square brackets, and the ovals with round parentheses. Therefore
the equivalent expression for Figure 5.1 in the linear form is as follows:
[concept 1](relation)[concept 2].
The reading of the expression in both forms can be: the relation of a
concept 1 is a concept 2. Sometimes the reading of an expression may
sound grammatically awkward. Consider the following example:
[Vehicle](part)[Trailer].
The reading of it, A part of a vehicle is a trailer, can be simplied as a
vehicle with a trailer. These readings are helpful mnemonics, not linguis-
tic rules for mapping the graphs into English sentences. Conceptual
graphs are nite, connected, bipartite graphs. Conceptual relations may
have any number of arcs; however, most of the common ones are dyadic.
Some are monadic, such as the past tense maker (PAST) or the negation
(NEG); others are triadic, like between (BETW); and any other number of
arcs.
The primitives of the conceptual graphs are concept types, concepts and
conceptual relations. The concept types represent classes of entity, attribute,
action, state and event. Examples of such concepts are: CAT, ANIMAL,
ARRIVE, ATTRIBUTE, BIG, COMMAND, COLOUR, JUSTICE,
LOVE, and WARM. Any conceptual graph system may have a set of types
as a basis. This set of concept types is organised using a relation _ dened
over the set. The concept types can be partially ordered according to the
super-type and sub-type relations. For example, consider concept types,
PHYSICAL-OBJECT, ANIMAL, MAMMAL and CAT, they can be
arranged as in the following order:
CAT _ MAMMAL _ ANIMAL _ PHYSICAL-OBJECT
which means that CAT is a kind of MAMMAL, every MAMMAL is a
kind of ANIMAL, etc. CAT is said to be a sub-type of MAMMAL,
MAMMAL a super-type of CAT. The type hierarchy is not a tree-like
structure, but a lattice structure. This can be seen from the organisation of
the following concept types:
ELEPHANT _ MAMMAL
ELEPHANT _ WILD-ANIMAL
RATTLESNAKE _ WILD-ANIMAL
TIGER _ MAMMAL
TIGER _ WILD-ANIMAL
Concepts are instances of concept types. For example, a concept the
mammal known as Clyde can be written in the conceptual graphs of
Polovina and Heaton (1992):
[MAMMAL: Clyde].
A concept that appears without an individual referent corresponds to a
generic referent. Such generic concepts could be denoted as
[Type_Label: *]. Writing [Type_Label] is merely a convenient
shorthand. Generic concepts may take up an individual referent (Clyde
would have a particular trunk). A unique number referent can suce to
make a concept distinct. Thus the generic concept [TRUNK] may become
[TRUNK:#1234] in respect of [MAMMAL: Clyde]. The following would
be produced:
[MAMMAL: Clyde] (part) [TRUNK: #1234].
Conceptual relations are links between concepts, and they indicate the role
that each concept (or percept, in terms of referent) plays. Typical examples
of conceptual relations found in the dictionary in Sowa (1984) are agent,
cause, duration, experiencer, instrument, location, negation, part, past, point-
in-time, successor, etc. Figure 5.2 shows an example of how a propositional
attitude think is handled in a conceptual graph. Other information on con-
texts, such as negation, tense, modality, and propositional attitudes can
also be incorporated in this way.
5.3 Some fundamental issues of information analysis
Any method of information analysis must adopt fundamental assumptions
in these four aspects, which one can always examine: ontological position,
person:
John
agent think
object
proposition:
cat agent sit location mat
Figure 5.2. An example of a conceptual graph (Jackman & Pavelin 1988).
epistemological position, social context and presentation (Hirschheim et al.
1995, p. 156).
The ontological position is concerned with the questions of the nature of
reality or of what is being studied. Is there an objective reality that is
independent of anyone? How does the world come to exist? What con-
stitutes the application domain that is being studied and modelled? Is
the object system (or the application domain) that is being studied
dened, created and sustained by physical components, operations,
roles, decisions, social actions, speech acts, or something else? These
are typical questions concerning ontological assumptions.
The epistemological position explains how people are to come to under-
stand the world, how to acquire knowledge about the application
domain. Is there an objective truth? How do we justify the knowledge
acquired about the application domain? How do we dene the validity
of knowledge and how do we prove it? The epistemological position a
method adopts is to a large extent determined by its ontological pos-
ition.
The assumption on the social context determines rst of all how to see
the relationship between information and action. Information is
created and processed by people and will serve people in their decision
making and actions. The value, relevance and validity of information
are all related to the social context in which information is used.
Secondly, a proper information analysis exercise can only be done if
the people involved use the same language and have a substantial
common understanding of the problem domain.
Finally, the assumption in respect of presentation determines the informa-
tion type to be presented and the form of presentation. For example,
some methods present information in a way that is close to the use of
natural language, while the relational data model recommends nor-
malised data entities (i.e. tables). Eciency of the presentation is also
part of the concern.
In addition, a sound method of information analysis must meet these crite-
ria.
Adequate A method of information analysis normally includes
identication of information requirements, followed by elicitation and
specication of requirements. Techniques for dening the scope of the
application domain for analysis should be available. In many cases,
prospective users of an information system may not know exactly
what they want. The information requirements are dicult to know
5.3 Some fundamental issues of information analysis 57
beforehand. The method, therefore, should assist people involved in
analysis to elicit the requirements systematically as the requirement
specication develops.
Instructive A method should not be like a cooking recipe; it should
provide guidelines and allow creativity. But it should provide clear
instructions on how to perform analysis step by step.
Capable of capturing rich business semantics An information model
should reect the patterns of behaviour and rules of the business
domain, which explicates the meaning of the application domain, or
business semantics. The business semantics should be clearly expressed
and treated as the focus of the modelling, not the technical design
issues (such as data formats for storage).
Rigorous An information model resulting from an analysis should be
precise in representing meaning. It should not lend itself to ambiguous
interpretation.
Easy to verify/falsify the validity of the information model An information
analysis is often a process involving a group of people with multiple
roles and perspectives. The prospective users, who are stakeholders,
will normally have to approve the result of analysis, based on the
outcome of validation of the information model. A sound model
should be possible for users to understand without having to go
through a lengthy training process in order to make sense of the model
comment on, critique and nally approve or disapprove the model.
Many methods in fact discourage users from checking the analysis
with complicated syntactical notation, technical jargon and imple-
mentation-related issues, and therefore they have to approve the analy-
sis without really understanding the information model which often
leads to a computer system that they are committed to.
These fundamental issues and criteria can be useful when one evaluates and
selects methods of information analysis for systems development or other
purposes.
5.4 The role of information analysis
Ultimately, an exercise of an information analysis addresses two issues.
First of all, it is concerned with understanding of the business itself and
business needs. Secondly, it is about how the needs can be met by providing
an organisational and technical system. So the information analysis deals
with both understanding of a problem and production of a solution. A
proper information analysis will enable system developers to ensure that
they address the right problem and answer the right questions. It should
then deliver the right information requirements on which a system solution
can be designed. Therefore an information analysis is extremely important
in the development of information systems.
Empirical studies show on many occasions an information system
project did not address the right questions because the requirements were
understood wrongly. Evidence also shows that an error at the information
analysis stage can be incredibly expensive. Figure 5.3 summarises these
ndings and shows the orders of magnitude of cost of errors at major
phases of an information systems lifecycle. From this diagram, one can see
that activities in the analysis and design phases have eects on the relevant
phases on the side of realisation. For example, the quality of the require-
ments analysis aects the running of the business. The design of an IT
system solution inuences the running of the IT system. The construction
and delivery of an IT system determine the installation and overall result of
the system implementation in the business organisation. An error at the
phase of requirements analysis can cost tremendously more than an error
of coding at the construction phase. Therefore, an adaptation of proper
methods and a careful analysis of information requirements are important
because they deal with the activities of the rst phase in systems analysis
and design.
An information analysis normally involves a group of participants and
covers a range of activities. These participants are called stakeholders who
all have an interest in the system development. The stakeholders are the
concerned parties who are either involved in the system development, the
owner of the system, or the user of the system (much literature can be
found in stakeholder analysis, e.g. Kolkman (1995)). The process of infor-
mation analysis carried out by the parties concerned is characterised by
these activities (after Kotonya & Sommerville (1998)).
5.4 The role of information analysis 59
Analysis & design phases Cost of errors Realisation phases
Requirements
analysis
10
6
to 10
9
Running the
business
Design
Running the
IT system
Construction Implementation
10
3
to 10
5
10
1
to 10
2
Figure 5.3. Cost of errors at dierent phases of information systems development
(after Liu et al. (1994)).
Requirements elicitation The information requirements and system
requirements are discovered through consultation with stakeholders,
from system documents, domain knowledge and market studies. Other
names for this process are requirements acquisition and requirements
discovery.
Requirements analysis and negotiation The requirements are analysed in
detail and dierent stakeholders negotiate to decide on which require-
ments are to be accepted. Stakeholders may have dierent require-
ments in terms of information and system functions. Conicts may
exist in these requirements. There are also possible ambiguities in the
requirements captured. The negotiation and analysis allow
clarication of the meaning of requirements. They also provide
opportunities for stakeholders to decide on the set of agreed require-
ments for the system.
Requirements documentation The agreed requirements are documented
at an appropriate level of detail. The requirements document should
be understandable by all stakeholders, so that all stakeholders can
check and validate the requirements (the objective of the next activ-
ity).
Requirements validation The requirements must be checked for consis-
tency and completeness. To achieve this, it would be essentially
benecial that all stakeholders can take part in the validation. The
involvement of stakeholders can boost the quality of the require-
ments, and hence improve the quality of the system. (However, this
might be dicult to achieve with many methods because of technical
complexity embedded in the presentation of the requirements.)
The information analysis and requirements engineering are not a straight-
forward process. It normally requires several iterations before a set of
requirements with a reasonable accuracy and condence can be produced.
The method of Semantic Analysis enables analysts, with the participation
of other stakeholders, to produce correct and accurate requirements in a
rigorous manner.
6
Semantic Analysis
A business system is a real information system, which is innitely rich and
complex. One can never exhaustively study a social or business organisa-
tion. What is modelled is always a part of it; and what is seen is just a reali-
sation (or an instance) of a pattern of behaviour of some agent. If one
wishes to model a business system one has to focus on the level of pattern of
behaviour. The method of Semantic Analysis provides a means for these
purposes by devising a canonical formalism with focus on the responsible
agent and his repertoire of behaviour. This method covers a range of activi-
ties in information modelling such as requirements elicitation, analysis,
specication and representation.
6.1 Theoretical aspects of Semantic Analysis
The relationship between behaviour and knowledge is in two directions. In
one direction, an agent learns about the world through his actions to gain
knowledge so that he gets to know exactly the meanings and boundaries of
objects, concepts and relations. On the other side, the behaviour manifests
knowledge that the agent possesses. The functions of both sides work at the
same time, i.e., what he knows tells him what to do; what he does reects
what he knows (partially, because what he does may be a kind of trail to
learn some new knowledge). Any rational action of the agent is constrained
by the repertoire of behaviour and directed by his knowledge of the world.
This repertoire of behaviour can be seen as aordances.
6.1.1 Aordances
Aordance is the word that Gibson (1968) and the direct perception
school of psychologists (see Michaels & Carello (1981)) use for behaviour
61
of an organism made available by some combined structure of organism
and its environment. The aordances of the environment are what it oers
the animal, what it provides or furnishes, either for good or ill, Gibson
(1968) writes. He gives an example to show that, if a terrestrial surface is
horizontal, at, suciently extended, and rm, then the surface aords
support. It is a surface of support. It is stand-on-able, permitting an upright
posture for quadrupeds and bipeds. It is therefore walk-on-able and run-
over-able. It is not sink-into-able like a surface of water in a swamp. The
properties of the surface, when they were measured as an aordance of
support for a species of animal, have to be measured relative to the animal,
not just treated as physical properties. According to ecology, a species of
animal is said to utilise or occupy a certain niche in the environment. The
natural environment oers many ways of life, and dierent animals have
dierent ways of life. The niche implies a kind of animal, and the animal
implies a kind of niche. They together constitute the totality of the world
that we inhabit.
The theory of aordances can be naturally extended to the social world
for studying social behaviour. The society acting as environment makes
many patterns of behaviour possible. People are agents living in a world of
ux. This world is socially constructed through agents actions, on the basis
of what is oered by the physical world itself. The agent-in-its-environment
makes it possible to acquire certain abilities of behaving which may be com-
pletely impossible if the agent is separated from the environment. While an
agent acts, he experiences the environment and alters it at the same time.
His previous actions build up a possibility of some new type of action. For
example, a persons ability to recognise a telephone is socially aorded by
the fact that there is an artefact called a telephone. Only if he has the ability
to recognise a telephone can he be taught how to operate it, and can he then
be able to make a telephone call. An agent learns his direct knowledge
through detecting valuable invariants by act-in-relation-to-environment
physically and mentally. By knowing the invariants, the values of certain
ranges of behaviour are learned. Having sucient experience in the social
environment and knowing the invariants enough, an agent can know the
boundaries of behaviour which are socially acceptable. However, what he
learns is only his direct environment, here and now. To be able to know any-
thing beyond here and now, signs must be employed so as to compose
propositions and images.
By using signs, another dimension of the world is opened up for the
agent, and he can then get access to the part of the world which is beyond
the here-and-now. Signs aord appropriate action at a distance. In particu-
lar, they aord coordinated actions among members of a society. People
62 Semantic Analysis
create signs and assign meanings to them, and use them to refer to the
future and past. This ambience, full of man-made signs, is called culture,
which supports most of the daily social interactions. A natural deduction,
which can be socially conrmed, is that dierent cultures, for example the
western or eastern, Christian or Buddhist, provide dierent aordances,
hence dierent perceptions of reality.
6.1.2 Ontology and some other fundamental notions
Lacey (1976) oers the following explanation about ontology:
A central part of metaphysics is ontology. This studies BEING, and in particular,
nowadays, what there is, e.g. material objects, minds, PERSONS, UNIVERSALS,
numbers, FACTS, etc . . . A particular theory about what exists, or a list of existents,
can be called an ontology.
1
Questions that an ontological theory tries to investigate are fundamental
and general. It seeks an inventory of kinds of things that exist and asks
what can be said about anything that exists, just in so far as it exists. Can we
classify all that exists into dierent fundamental kinds (categories), in one
way or more ways? Is there any hierarchy among kinds of things? Do some
depend on others for their existence? These are typical questions concerned
in any ontological study, and involve the relations between very general
notions, such as thing, entity, object, individual, universal, particular, sub-
stance, and also event, process, state.
In a paper, Stamper (1985) examines three very dierent ontologies
which have strong inuences on the studies of information and knowledge.
The rst one only recognises symbol structures; this is the ontology of a
relational database language. The second ontology recognises distinct, indi-
vidual, identiable objects; this lies behind predicate logic as its formal
semantics is based on set theory. The third ontology assumes that the only
thing we can know is our own behaviour in our own environment; this is the
ontology that Semantic Analysis bases itself on. This ontology assumes
that the world known to a particular agent comprises only the actions he
can perform in his environment.
The agent can be as simple as an individual person, and complex as a cul-
tural group, language community or society. What exists in the world relies
on the repertoire of behaviour of the agent. For example, an ink pen may
exist in a sense of an ink pen in one cultural community, but of an unknown
object in some other cultural community; a particular sword may exist as a
6.1 Theoretical aspects of Semantic Analysis 63
1
The capitals used by Lacey indicate his view that some of these categories are more impor-
tant than others.
weapon in one society but as only an ornament in another. In these exam-
ples, the existences of the pen and the sword depend on the agent whose
aordances allow him to recognise them, to assign meanings to them, to
know their properties. This leads to the notion of dependency in this ontol-
ogy: the agent learns abilities; these aordances enable him to act, in ways
that determine the shape of the world he inhabits. The aordances the
agent has gained are universal invariants which constitute the repertoire of
his behaviour, and allow him to react recurrently and circumstantially.
Every particular action he performs is a realisation of a universal pattern of
behaviour. The two main categories of the construction blocks of the world
are agent and aordance, but the denition of agent and aordance is
context-sensitive. Some simple agent, for example a person, is ontologically
dependent on some other complex agent, say society; in this case, the agent,
person, is an aordance itself. Nevertheless, a criterion for determining
whether an aordance can ever be an agent is that an agent must be able to
act responsibly. Properties of the agents and aordances are another fun-
damental class of the construction blocks, labelled as determiners, which
result from complicated measurements of the agents and aordances.
Actions of agents normally should be modelled at the universal level
concerning the universal patterns of behaviour in knowledge representa-
tion and requirement analysis, though in the end they will be recorded in a
database at both levels. Loux (1978) reports that Aristotles denition of
the universal is that which is of such a nature as to be predicated of many
objects. Frege considers that concepts are the referents of predicate-
expressions. Russell states that substantives, adjectives, prepositions, and
verbs, the expressions which serve as predicate-terms, stand for univer-
sals. In NORMA, the universals represent invariant patterns of behaviour,
or mechanisms. The universals can have realisations which are particular
instances. This universalparticular conceptual construct is found in con-
ceptual graphs as well (see Sowa (1984)). Universals are represented as the
concept types, while particulars are concepts. However, the representations
of both universals and particulars in NORMA are done in the same way,
because there is no fundamental dierence between the universals and par-
ticulars. Both of them represent repertoires of behaviour. Similar treat-
ments of both universals and particulars can be found in Semantic Analysis
in later sections.
6.2 NORMA
NORMA is a knowledge representation language developed in the
MEASUR research programme (Stamper 1992, Stamper et al. 1988). A
primary function of this language is for conducting semantic analysis.
NORMAcomes from NORMand Aordance, and is devised as a language
for specifying norms and aordances as systems analysis and requirement
specication. The purpose of this section is to introduce NORMA in terms
of some of its most important concepts and syntax.
6.2.1 Well-formed formula
The focus of systems analysis is the agents in action. No matter how
complex the agents are in a social environment, they can be studied by
observing their actions and detecting the invariants, i.e. general patterns of
behaviour. Two axioms stated earlier, in Chapter 2, that dominate basic for-
mulation of the language are
there is no knowledge without a knower, and
there is no knowing without action.
These axiomatic, philosophical assumptions suggest that the only feasible
way to depict any fragment of knowledge is to model the agent and his
action at the same time. The meaning of representation of action is only
determined if the viewpoint is chosen; therefore, the place of the agent in a
knowledge representation model is essential. Following this, a well-formed
formula (w) for knowledge representation is derived as always having the
following structure:
r knower-termbehaviour-term,
or
agent-term action-term.
For example,
Ax, where A is an agent and x stands for a pattern of action, e.g. person
stand.
This w requires every piece of knowledge to be tied to the agent who is
responsible for it. The knowledge can be about universal patterns of actions
or particular instances. In any case, the agent is responsible for possessing
the knowledge, performing the action, or designating meanings to actions.
6.2.2 Aordance and ontological dependency
Some aordances can be possible only if certain other aordances are
available. This kind of dependency is ontological, which is taken as an
essential theme in representing knowledge. For example, for a person to be
able to stumble, he must rst walk; for two persons to be separated, they
6.2 NORMA 65
must have a marriage or other relationship which involves being together.
In these two examples, walk and marriage are ontological antecedents for
stumble and separation respectively which are called ontological depen-
dants. These ontological antecedents are more direct extensions of
aordances of the agent or agents. An agent with aordances can be seen as
a modied agent, the aordances being treated as extensions. To represent
the above two examples, the following expressions can be obtained:
person walk stumble, where person walk is the modied agent
which aords stumble; and
(person
1
, person
2
) marriage where (person
1
, person
2
) is a
separation, complex agent, the modied agent is
(person
1
, person
2
) marriage on which a
further aordance separation is an
ontological dependent.
These two expressions have employed syntactic formulae:
Axy, e.g. person walk stumble; and
(A,B)xy, e.g. (person
1
, person
2
) marriage separation.
In most of the cases, analysis of agent (either a simple one, such as a person,
or a complex one, such as a corporate body or an institution) with his
behaviour would be a recursive application of this principle of modied
agent to obtain the expressions in the w.
The aordance can be composite. When an agent is doing one thing, he
might also be doing another. Thus,
A(x,y), where , reads as while, a restriction. E.g. person(stand, speak),
meaning: a person stands while he speaks.
The agent may do one or another or both of two things at a time. The
syntax for this is:
A(x:y), where : reads as or-while, a conation. E.g. person(speak:
sing) for a person either speaks or sings.
If the agent has an aordance of doing one thing while the other is not
experienced, the following syntax is designed for designating that:
A(x;y), where ; reads as while-not, an exclusion. E.g. person(sit; sing)
for a person sits while he is not singing.
A composite aordance can be partitioned into several parts. The notation
in NORMA is dened as follows:
Ax.y, which means y is a part of x. E.g. university student-council.
chairperson, the student-council is ontologically dependent on the
university, and the chairperson (the role or incumbency in the oce
rather than the person) is a part of the student-council.
6.2.3 Semiotic behaviour
Two types of behaviour have to be identied: substantive and semiotic. An
agent can performan action and he can also use signs to describe the action
(though the using of signs itself is an action). Without involving signs, the
agent is conned within his here-and-now environment, for example, he
stands, walks, and moves about. If he wants to make his actions linked with
the past and future, he must use signs. The abilities to use signs in various
ways are called semiotic aordances, while the behaviour directly bringing
in a change of a physical or social world is considered substantive. For
example, marriage can be considered as a substantive behaviour of the two
persons concerned. The associated semiotic behaviour with marriage
would be announcing the start or the nish of the marriage where produc-
tion of semiotic entities (signs), such as speeches and documents, is
involved. Such semiotic behaviour can be represented in the following form:
B Ax, where B is the agent performing the semiotic action, and between
and including the quotes are signs employed by B in his behaviour. Ax is
the meaning or referent of the semiotic action. E.g. Mary John happy.
6.2.4 Time
Time is one of the most eective measures to determine the meanings of
concepts, actions, and events expressed in our language. Aordances have
an ontological nature which allows them to be placed into the ux of the
world in terms of time. The benet of using time is exploited in NORMA in
many ways. When an agent and an aordance are identied and dened,
their existence period must be determined, marked by start and nish, for
both a universal and a particular. The determination of the boundaries in
terms of time is the minimum as a guarantee that the meanings of the
words and expressions appearing in an analysis are clear. It is also used as a
clue for seeking agreed meanings from dierent viewpoints. Several opera-
tors are dened in NORMA to enable one to describe patterns of behav-
iour that are related to the start and the nish of the aordances. Two of
them are beginning and ending:
6.2 NORMA 67
Ax, e.g. person stand, meaning: a person begins to stand, which may
terminate an action of sitting and leads to a start of standing, if the
beginning process is successful.
Ax, e.g. person stand, meaning: a person ends standing, which may
bring the action of standing to a nish, if the ending process is successful.
6.2.5 Determiner and identity
Agents and aordances have properties which are invariants of quality and
quantity that dierentiate one instance from another. This kind of invari-
ant is called a determiner. Typical examples of determiners are names,
addresses, weights and hair-colours of a person. To describe the determin-
ers, the following mechanism is devised:
A#x, or Ax#y where the hash sign indicates what follows is a determiner,
e.g. person#name, person#weight, or person salary#amount, etc.
An agent may have many roles to play when he is involved in many actions
or relations. For example, a person can be seen as a father, a husband, a
manager, and a customer, depending on the circumstances in which the
person is looked at. The role name, essentially serving the same purpose as
the other properties discussed above, is also a determiner of individuality.
See the following expression as an example:
(person#husband, person#wife)marriage
This describes that a marriage is a joint aordance of two persons, of
whom one has a role name husband and the other wife.
6.2.6 Genericspecic relationship
Agents and aordances can be placed in genericspecic structures accord-
ing to whether or not they possess shared or dierent properties. The
genericspecic classication is normally determined by norms which may
be socially or culturally formed. In some cases, the classication rules can
be dened by the parties involved in the particular practical aairs. The
formula to represent the genericspecic relationships is as follows:
A((b:c:d:e) f), where b, c, d, and e are specic aordances of the
generic f. E.g. society ((natural-person:corporate-
body:government) legal-person), which means in a
society, there are several specic kinds of legal persons,
such as natural persons, corporate bodies and
governments.
The genericspecic structures are contingent on social decisions and are
not logically necessary. For example, one may classify a whale into the cate-
gory of sh or mammal, or both.
6.2.7 Dening authority and responsibility
The notion of responsibility is as essential in NORMA as the notion of
truth in classical logic, because in NORMA any recognition of aordances,
and their existence periods, are tied to the agents. Realisations of actions
and changes of states are all decided by the responsible agents in accor-
dance with the authorities. For example, the commencement and termina-
tion of a studentship are authorised by someone who is in a role of student
registration ocer; and the qualication of enjoying a social security
benet is determined by the special government oce. The responsibility
and authority over realising each instance of the aordances are analysed
and traced in the analysis. The following way is dened for denoting
authority in NORMA:
A(x@ y) which means the authority for x is y. E.g. nation(law@
parliament).
6.2.8 Graphic representation ontology chart
An old Chinese saying says A picture is worth a thousand words.
NORMA oers a way of graphic representation. Agents and aordances
are nodes in the graphic representation, and linked with lines. The
antecedents are placed on the left of the dependants. In other words, the
leftright positioning reects the ontological dependencies. The example in
Figure 6.1 explains how the ontological relationships can be represented in
the graphic form of NORMA.
The society is the root agent in this model, and it has two aordances:
person and thing. Here person is an agent and thing is a normal
aordance. Both are ontologically dependent on the society, which means
both are dened in the context of a certain society. If the society does not
exist, then the concepts of person and thing become undened. Therefore,
it is the society which has aordances of recognising the person and
thing and the other dependants. If the society is seen as a modied agent
with the extension of aordances of person and thing, it makes it possible
6.2 NORMA 69
to talk about owns. More straightforwardly, a person can own a thing.
Agents are placed in ovals, such as society and person. Role names are
put in half-curves, e.g., owner, seller and buyer. Actions and other kinds of
aordances are put in rectangular boxes. Notice the action sells is ontolog-
ically dependent on the owner (the role) and owns, and the action buys is
built upon person and owns. This suggests that selling is only possible for
the owner who owns the thing, while buying is for any person. Selling and
buying are referred to the aordance owns. That means when people are
trading, it is the ownership rather than the physical thing itself that is dealt
with. In this sense, the representation ontologically reects the social prac-
tice which is dominated by the shared norms in the social context.
The example in Figure 6.2 involves the representation of aordances
having genericspecic relationships. The root agent nation aords legal
person within the given social and legal context. Under the generic
heading, more specic items can be found: natural person, and corporate
body. Within the heading of corporate body, there are more specics:
items company, government, and learned society. The specic kinds of
aordances under a generic heading share some properties; for example, all
specic kinds of corporate body have a management as part of the organi-
sation. Any natural person can be employed by a corporate body, in which
he is called employee and the other party is employer. The management is
a part of the corporate body. Therefore a line with a dot associated is
drawn to indicate the wholepart relationship between them. To be a
buyer
o
w
n
e
r
seller
person
society
thing
buys
sells
owns
Figure 6.1. An illustration of NORMA in graphics.
manager, one is rst of all considered as an employee of the corporate
body, which is represented by putting being an employee as an ontological
antecedent for the incumbent. This constraint is assumed, when the analy-
sis is conducted, as a norm generally residing in the social organisations.
The ontological relationship is considered as the most fundamental rela-
tionship to model in NORMA. All other relationships, such as
genericspecic, wholepart, determiners, roles, etc., are treated on the
basis of the ontological relationship. Therefore, the process of producing a
graphic representation in NORMA is called ontology charting, and a com-
plete conceptual model represented in NORMA graphic notation is called
an ontology chart. Table 6.1 presents some of the syntax notations of
NORMA in both textual and graphic forms, and summarises briey some
of the discussions in this section.
6.3 Using LEGOL to specify norms
LEGOL is a knowledge manipulation language that can be used for han-
dling data that have rich semantic and temporal features. In this section,
some LEGOL operators will be used even before the language itself is
introduced in Chapter 10.
The ontological structure of the agents and the aordance depicts the
most fundamental norms of behaviour. The complex behaviour which is
built up on the fundamental patterns of behaviour can be described in the
form of norms. In addition, although the ontological structure oers an
understanding of what patterns of behaviour are ontologically available,
the detailed conditions and constraints of the realisations of the behaviour
6.3 Using LEGOL to specify norms 71
e
m
p
lo
y
e
r
natural person
corporate body
company
government
learned society
nation
m
a
n
a
g
e
r
incumbent
management
employs
legal person
e
m
p
l
o
y
e
e
Figure 6.2. Another illustration of NORMA in graphics.
T
a
b
l
e

6
.
1
.
P
a
r
t

o
f
N
O
R
M
A

s
y
n
t
a
x

a
n
d

e
x
a
m
p
l
e
s
.
T
e
x
t
u
a
l
G
r
a
p
h
i
c
M
e
a
n
i
n
g
E
x
a
m
p
l
e

i
n

g
r
a
p
h
i
c

r
e
p
r
e
s
e
n
t
a
t
i
o
n
A
x

A

x

x

i
s

a
n

a
o
r
d
a
n
c
e

o
f
A
p
e
r
s
o
n

w
a
l
k
s
o
c
i
e
t
y

p
e
r
s
o
n
.
.
A
x
.
y
A

y
y

i
s

a

p
a
r
t

o
f
x
;
t
h
e
y

a
l
l

a
r
e

a
o
r
d
e
d

b
y

A
n
a
t
i
o
n

s
c
h
o
o
l

d
e
p
a
r
t
m
e
n
t
A
x
#
y

A

#
y
A

a
o
r
d
s

x

w
h
i
c
h

h
a
s

a

d
e
t
e
r
m
i
n
e
r

y
s
o
c
i
e
t
y
a
,
b
,
c
,
d

a
r
e

s
p
e
c
i
c
s

o
f
f
n
a
t
i
o
n
A
(
(
a
:
b
:
c
:
d
)
f
)

A
(
A
#
x
,
B
#
y
)
z
A

w
i
t
h

r
o
l
e

n
a
m
e

x
,
a
n
d

B

w
i
t
h

r
o
l
e

n
a
m
e

y

j
o
i
n
t
l
y

a
o
r
d

z

(
e
m
p
l
o
y
e
r
)
c
o
r
p
o
r
a
t
e
-
b
o
d
y
e
m
p
l
o
y
s
(
e
m
p
l
o
y
e
e
)
p
e
r
s
o
n
l
e
g
a
l

p
e
r
s
o
n
p
e
r
s
o
n
c
o
r
p
o
r
a
t
e
-
b
o
d
y
p
e
r
s
o
n
#
w
e
i
g
h
t
(
x
)
A
z
(
y
)
B
f
a b c d
patterns are not covered. Norms are needed for specifying conditions
and constraints of the actions as well. Norms have the following general
structure:
conditionconsequent.
Norms exist in a community. Rules are signs for norms. At the analysis
stage, the norms can be identied and specied in natural language, but
complex or subtle norms may not be capable of representation in formal
rules. Specied norms can be used to understand the organisations as well
as to be programmed for automatic execution. The following are some
examples of norms specied in LEGOL. These examples are based on the
IFIP case study on information systems (Olle et al. 1982), and a complete
semantic model and more detailed illustration of LEGOL are presented in
Appendix B.
contributor(paper) author(paper)
contributor(report) referee(paper)
These two norms dene the roles of author and referee.
member(WG or-while TC) eligibility#priority#1
author(paper#selected) eligibility#priority#1
These two norms dene the conditions for assigning the rst priority
to the applicants of the conference. A person who is a member of
the working group (WG) or the technical committee (TC) is assigned
the rst priority, and the author of a selected paper is assigned the
rst priority (boldface words appearing in the expressions are
LEGOL operators).
6 month before start-of meeting#CRIS-2
while (selected(paper) while-not invited(author(paper)))
print author
This is a trigger norm for actions, a reminder for the IFIP CRIS-2
conference secretary: a check to be done half a year before the
conference. If there is any author of a selected paper who has not
been invited, his name will be printed (then, the secretary can send
him an invitation to the meeting).
6.4 Conducting a Semantic Analysis
Semantic Analysis is a method for eliciting and specifying user require-
ments. One of its theoretical foundations is organisational semiotics
(Stamper et al. 1997) and the semiotic framework. The method has been
applied in many elds such as user requirements for business systems,
organisational analysis, legal document design, and analysis and design for
computer systems (Ades 1989, Liu 1993, Liu et al. 1994, Liu &Dix 1997).
The method of Semantic Analysis can be summarised in a few major
phases as schematically presented in Figure 6.3. The rst phase is for people
involved to receive a document which is sometimes called a problem state-
ment or denition, and for them to understand the problem. The people
involved may be a single analyst or, for a large project, often a group of ana-
lysts with users participation. The next phase, the candidate aordance gen-
eration, is to produce a list of vocabulary of semantic units that may
possibly be used in the semantic model to describe agents and their patterns
of behaviour. The phase of candidate grouping will allow further analysis of
the semantic units. Following that, there is ontology charting, where a com-
plete semantic model is produced. Finally, Norm Analysis, as a separate
method, can be carried out to identity norms, and to link the norms with
each part of the semantic model. Below, each of these phases will be elabo-
rated. A case study of project management
2
will be used to illustrate the
whole process of Semantic Analysis.
6.4.1 Understand the problem domain
A Semantic Analysis exercise always begins with the phase of problem
denition. In this phase, the assignment of system development is given in
the form of a written document where the problem is dened. But the
written ocial denition would normally not be enough to capture a com-
2
After a case study used in a series of lectures and workshops run by the MEASUR team in
Utrecht and Twente between 1991 and 1993.
problem definition
candidate affordance
generation
candidate grouping
ontology charting
Figure 6.3. Major phases of Semantic Analysis.
plete picture of the problem. Therefore, thorough investigations have to be
done by studying the relevant documents. In addition, interviews with
potential users of the envisaged system can be carried out for collecting
supplementary information.
The following is the problem statement of project management with some
simplication, which describes the case study that is going to be used to
illustrate how to conduct Semantic Analysis. Our task here is to take the
problem statement as the basis for producing a semantic model using
Semantic Analysis. The result can be used either for understanding the
mechanism and practice of project management in the company, or as a
basis for designing a computer system for project management purposes.
In the company, departments are responsible for projects. Each depart-
ment and each project has a budget.
Employees work in one department but can be assigned to dierent pro-
jects. Projects do not belong to a single department. Work activities are
charged at an hourly rate which depends on the function of the
employee.
The total time which an employee spends on a project is imputed to it at
the hourly rate, and the total cost of the project is computed.
The problem is basically introduced in the problem statement, so by
reading it the analyst can gain an overview of the problem. However, some-
times the problem statement may be too vague and the requirements may be
fuzzy. In this case, gathering more information about the problem is
needed, which is normally done by collecting relevant documents and con-
ducting interviews with the problem-owners. Those documents and inter-
views, as important supplements to the formal denition, can help the
analyst to understand better the words and expressions in the problem
statement.
6.4.2 Generating candidate aordances
The second phase, candidate aordance generation, is to study the collected
papers and notes, and to single out the semantic units, for example nouns,
Table 6.2. Candidate aordances.
company charged department hourly rate
responsible for depend on project function
budget total time employee spend
work imputed to assigned to total cost
belong to computed activity
noun-phrases, verbs, verb-phrases, and prepositions, which may indicate
possible agents, aordances and other relationships.
In this phase, semantic units are identied from the problem denition as
listed in Table 6.2. Every word in the problem denition is in principle
useful in analysis; therefore, one has to be careful when a word is going to be
ignored. Even an article, a preposition or an auxiliary verb may suggest
some additional information to a concept, a relationship, or an intention,
though nouns and verbs are more likely to suggest concepts of agent and
aordance. The words and phrases that form semantic units are singled out
as candidate aordances, and they contain complete concepts. It can be
noted that some words are left out but that does not mean they will not be
included in the model. Some words are taken in dierent forms, e.g. from
the plural form of nouns to the singular form.
6.4.3 Candidate grouping
The phase of candidate grouping consists of a few small steps. First of all,
the semantic candidates can be categorised as agents, or aordances.
Among aordances, they can be further classied into universal action-
types, determiners, role names, and so on. The specic aordances should
Table 6.3. Candidate grouping (I).
company agent
department agent, a part of a company
responsible for aordance (a department)
project aordance
budget aordance
amount determiner of budget (required to add to quantity a budget)
employee role name (of a person who works in a company)
work aordance
assigned to aordance
belong to aordance (of department and project)
activity aordance (of employee)
charged aordance
hourly rate determiner (of function or activity?)
depend on aordance (of function?)
function determiner (of employee)
total time a derived value as determiner, will not appear in the model
spend aordance (of employee and project)
imputed to complex action governed by rules, will capture as norms
total cost a derived value as determiner, will not appear in the model
computed complex action governed by rules, will capture as norms
be grouped into generic families. Dependants will be connected to their
antecedents. The work in this step can be seen as a preparation for the next
phase.
In this phase, the functions of the semantic units listed are rstly studied
by labelling them as agents, aordances, role names, and determiners. They
are also classied according to the universalparticular and generic
specic relationships. As shown in Table 6.3, not only are these semantic
units classied, but at the same time additional information for explaining
the functions is noted, which will be useful for the next step in construction
of the whole ontology chart. In the table, one can see that functions of some
semantic units are not certain at this stage. For example, the hourly rate is
noted as determiner (of function or activity?), showing an uncertainty
on which antecedent the semantic unit, hourly rate, will have. Would it be
function or activity? A few more aordances may be added to the previ-
ous list, as they are needed for a sounder representation of the underlying
ontologies. Finally, sketches of piecemeal ontological structures are con-
ceived at this stage (see Figure 6.4).
These unconnected pieces of ontology charts help one to see better the
eect of the grouping of the semantic units. One can start with the easier
pieces, for example an organisation and a person jointly aord a depen-
dency employs, in which the organisation plays the role of employer and
e
m
p
lo
y
e
r
#function #hourly rate
permanent
contractor
organisation
person
organisation
department
employee
task
employs
project
e
m
p
lo
y
e
e
Figure 6.4. Candidate grouping (II).
the person plays the role of employee. Additional information may be
added at this time, such as to categorise the employees into permanent sta
and contractors, on a justication that this information is relevant in this
project management system. The piece of chart that has an organisation, a
department, a project, and so on, contains most of the information pro-
duced in the candidate grouping. For example, the department is a part of
the organisation. But in the early analysis, though it was identied that the
project is an aordance, it was not clear about its antecedent (supposedly
owing to a not very careful analysis). Therefore, the question here is: should
the project be connected to the organisation or the department? If one
reads the problem statement again, it can be seen that the project should
not belong to any department. Hence, the project is a dependant of the
organisation. It has also been found out that there should be two budgets,
one for each project and one for each department, as denoted in Figure 6.5
below.
An employee, as seen in Figure 6.4, has a determiner, function, which
itself has a determiner, hourly rate. Determiners are attributes that enable
one to describe an agent or aordance. For example, one can describe an
employee with his or her function which can be further described with an
hourly rate of how much he or she earns per hour. More such pieces can be
produced for discussion and verication with the users before a complete
ontology chart is produced.
6.4.4 Ontology charting
After the candidate aordances are analysed and grouped into scattered
pieces of ontological structures, the fragments will be assembled into an
integrated ontology chart. The ontological dependency between the frag-
ments is the key to their connections, and it is constantly checked and main-
tained. It is unavoidable that some mistakes in the last phase may have been
made and they will be corrected at this step. Figure 6.5 is the ontology chart
for the project management.
The model exposes a few basic assumptions underpinned by the ontolog-
ical principles. There is a root agent in the chart which functions as the ulti-
mate antecedent for the whole problem domain under study. This root
agent is the social community in which all its members share some funda-
mental concepts and cultures. Without these fundamentals, there would
not be agreed meanings of organisation and person. Dierent social com-
munities may have completely dierent denitions of an organisation.
Dierences may even exist in determining the boundaries of a person for its
start and nish, which may be the causes of debates on moral issues relating
to abortion and euthanasia.
A project is ontologically dependent on an organisation, but not a
department. This suggests that a project exists independently from a
department. The relationship between a department and the project is by
the departments participation. An employee working in a department may
or may not be assigned to a project. An employee who is assigned to a
project may work on a certain task, but the actual eort put into the task is
measured by the determined time spent.
This ontology chart must be read many times by the analysts with refer-
ence to the problem statement and other business knowledge that is gained
through various channels. After the draft of an ontology chart is produced,
checks and verications with input from the users are strongly suggested for
maximum support and assurance from them.
6.4.5 Norm Analysis
The analysis can then proceed to the application of the method of Norm
Analysis where norms will be identied and specied as the conditions and
constraints for realisations of the aordances. Norms specied here are
kinds other than the ontological dependency imposed on the antecedent
#function #hourly rate
#time
spent
person
e
m
p
l
o
y
e
e
society
organisation
e
m
p
l
o
y
e
r
employs works works on
assigned to
project
department
responsible for
task
#budget
#budget
Figure 6.5. Ontology chart for project management.
and dependant. In the case of the project management, everyone working
on a task must in the rst place be assigned to the project. This can be
specied in LEGOL:
assigned_to(employee, project) while task(project) works_on(employee,
task).
A norm has a validity period of time marked by a start and a nish. Norms
are, for example, institutional rules such as the age limit for employees, the
maximum number of projects an employee is allowed to participate in, and
total time spent on a project by an employee. Therefore, the validity time
during which the norm can be applied must be identied. All the norms are
linked to the relevant aordances as constraints. For example, the age limit
for employees is linked with the aordance employs; and the maximum
number of projects for an employee to participate in is linked to the
aordance assign to. Note that normally the norms should be linked to
aordances, the patterns of possible actions, rather than agents, because an
agent may have a number of patterns of possible actions, each following a
set of norms specic to that aordance.
A detailed discussion of the method of Norm Analysis is found in the
next chapter.
6.5 Commentary on Semantic Analysis
In an actual practice of Semantic Analysis, the order of the phases is not
strictly followed. For example, even when it is in the phase of ontology
charting, some new aordance may be identied, and it may then be added
to the model. Iterations of the analysis phases may be necessary in order to
reach an adequate analysis. Moreover, interactions between the analysts
and the problem-owners are essential during the analysis. In this way, opin-
ions and judgements can be inspired by the preliminary results, and com-
prehensive requirements can be elicited and incorporated into the model
well before any system is developed.
There are a number of special features of Semantic Analysis worth
noting. The rst one is the role of the ontological relationship in modelling.
The method always requires one to begin a Semantic Analysis by identify-
ing agents who create the social world around them. An agent has some
abilities to accomplish things. These abilities are also identied and put into
the analysis model to construct the picture of the world. In doing so, a rig-
orous analytical principle must be observed, i.e., the existence constraint, or
in other words, the ontological constraint. The ontological constraint says
that a pattern of behaviour can be described only if the agent who acts is
described in the model. So the agent is a kind of ontological antecedent to
his actions (the latter being the dependants of the agent). The ontological
constraint must be fully observed. That means that the existence of the
dependants must be within (or during) the existence of the antecedent. If
there are more aordances available when a certain behaviour is realised,
they should also conrm the ontological constraint. That is, any added
aordance should not exceed the existence of its antecedent aordances
already described in the model.
Besides the ontological constraints as a principle for analysis, there is
another important principle: it is not allowed for an analyst to invent
articial terms or introduce new concepts when modelling the agents
actions. The purpose of this is to force the analyst to speak the same lan-
guage as the problem-owners. Any ambiguity in the terms or concepts used
in describing the problem should be resolved by putting them into a context
of actions which are already described and understood. When doing so, if
the problem-owners are inspired with some new terms, they may be used
only after a careful justication by the problem-owners and the analyst.
The reason for this is that the world to be modelled is constructed by the
community of agents, i.e. the problem owners. The agents know the mean-
ings of words in their own world, their interpretations are the only ones
justied.
Another feature is the scope of the analysis. Jayaratna (1986, 1990)
observes that some methods of systems analysis have a narrow scope of
analysis; some issues of enquiry, such as problem-owners notional systems
(i.e. basic assumptions of the focal problem and the environment), tend to
be ignored. Some predominantly standard systems development
approaches take the physical world as an entry. For example, with Jacksons
development method, a physical world is given as consisting of objects in
motion from state to state; the task of the systems worker is to build a
model which is an abstract description of the real world (Jackson 1983).
Semantic Analysis, however, covers both the wet and dry aspects of the
problem (Goguens terminology 1992). It places the problem-owners at the
kernel of the total enterprise of analysis. The role of an analyst is only to
elicit and specify the business knowledge. The analysis is aimed at produc-
ing precise requirement specications which are intelligible to the users (Liu
et al. 1994), so the responsibility still remains with the users.
6.5 Commentary on Semantic Analysis 81
7
Pragmatics and communication
A sign has a meaning, and it can be used intentionally for communication.
Pragmatics, as a branch of semiotics, is concerned with the relationships
between an intentional use of a sign and the resulting behaviour of respon-
sible agents in a social context. Communication takes place successfully
when a meaningful sign is used with an appropriate intention between the
two parties involved (e.g. speaker and listener).
This chapter addresses the eects of using signs in organisations. Before
introducing the semiotic approach, various approaches to and techniques
for analysis of the pragmatic aspect of signs are discussed in the chapter,
such as speech act theory, functional grammar, and deontic logic. The semi-
otic approach distinguishes a number of elements in a communication act:
propositions, propositional (or illocutionary) attitudes and perlocutionary
eects. A communication act will further result in a variety of commit-
ments and responsibilities. The method of Norm Analysis introduced in
this chapter provides a formalism for one to identify and specify these
important notions in analysing and modelling social and organisational
aspects of communications.
7.1 Human communication
A social community is a purposeful system in which human beings act and
interact with each other for achieving goals that may relate to the commu-
nity or its individuals. Doing things with words in a concerted way
becomes a necessity for fullling ever more complex objectives of social
beings in ever more complex organisations. Communication is the indis-
pensable means by which agents in society coordinate their actions to
achieve their goals.
Alanguageactionviewof informationsystems leads one todesigninfor-
82
mation systems as part of the relevant human communication system;
analysis of communication, therefore, is the essential entry for information
systems analysis and design.
1
This school sees the uses of language as
actions. A radical recognition about language and existence Nothing
exists except through language, stated by Winograd and Flores (1987) is a
typical representation of this point of view. In addition to this view of lan-
guage as actions, another complementary viewshould be advocated, that is,
the notion of actions as language. Actions should be regarded as a kind of
language because they manifest ideas, thoughts and propositions. Using
non-linguistic means, such as facial expressions, gestures, and other osten-
sive actions, can also perform communication of messages and intentions.
An integrated languageaction viewhas been adopted here for information
analysis in the current project. According to this view, speech acts have the
capacitynot onlytorepresent the relevant social world, includingits actions,
but also to constitute part of its (social) actions. Accordingly, information
communicated through language has its complement in the social world of
situations, norms, institutions and cultural patterns; and vice versa.
Theories and formal methods are available for studying language and the
use of language in human communication. Speech Act Theory and
Functional Grammar provide facilities to analyse the taxonomy and func-
tionality of communication acts which will be discussed in this chapter. In
order to analyse the social eects of communication acts, deontic logic will
be examined and used to link the linguistic and the social acts. As com-
puter-based systems are introduced into business practice, how to utilise
this modern technology to enhance human performance in terms of inter-
action and communication becomes an important question. In order to
answer this question, the modelling of human communication is done in
this chapter by analysing communication acts employed and propositional
attitudes expressed. This analysis serves as a theoretical basis for the devel-
opment of a technical information system (e.g. Normbase).
7.2 Other approaches to communication
7.2.1 Speech Act Theory
Speech Act Theory originated in Austins 1955 William James Lectures at
Harvard (Austin 1980) and was further developed by Searle (1969). It
1
Some examples of such a languageaction approach can be found in Andersen (1991),
Auramki et al. (1988), Dietz & Widdershoven (1991), Flores et al. (1988), Lyytinen &
Lehtinen (1986), Weigand (1990).
studies roles of language use in communication and focuses on the perfor-
mative aspects of language. Searle proposed a classication of four sorts of
speech acts: utterance acts, illocutionary acts, propositional acts and per-
locutionary acts. An utterance act is simply the act of uttering an expres-
sion. An illocutionary act is a basic, meaningful unit of human
communication which consists of propositional contents and carries inten-
tions to be perceived by a hearer. A propositional act is a subsidiary act of
an illocutionary act. It expresses a propositional content which is realised
by uttering the expression. The eects produced on a hearer by an illocu-
tionary act are called perlocutionary eects and the acts of producing them
are called perlocutionary acts. Perlocutionary acts, unlike illocutionary
acts, are not essentially linguistic, for it is possible to achieve perlocutionary
eects without performing any utterance at all.
For the purpose of studying roles of a language in communication, the
most relevant acts are illocutionary acts and perlocutionary acts. In the
course of a conversation, a complete unit would always be an illocutionary
act which is produced by the speaker with certain meanings and intentions
incorporated in the utterance. The hearer receives this communication unit
and perceives the meaning from his own perspective which may result in a
dierent one from the speakers. In any case, there may be (or may not be, if
the hearer ignores it) eects on the hearer. The eects are in the mental
world of the hearer; e.g. he changes his feelings, mental states, etc. In this
case, thus, the perlocutionary acts are not overt linguistic acts and not
observable. He may also make explicit reactions to the perlocutionary act;
e.g. he accepts an invitation and changes his agenda, which is observable.
Figure 7.1 illustrates two agents involved in a process of communication to
show that the roles of the speaker and the hearer will change in turn. The
speaker addresses the hearer with an illocutionary act which produces
eects on the hearer; one should be aware that the eect on the hearer may
or may not be desirable from the speakers perspective. However, in a broad
84 Pragmatics and communication
speaker/
hearer
hearer/
speaker
Figure 7.1. Speaker and hearer. Solid line, illocutionary act; dotted line, perlocu-
tionary act.
sense, the speaker himself is a hearer while he performs the utterance; there-
fore, there is also a perlocutionary eect on him. The perlocutions are
directed to the internal perlocutionary eects on the hearer (Eemeren &
Grootendorst 1984). However, consequent behaviour is observable and
that indicates the perlocutionary eect. In the above example of issuing an
invitation to someone the perlocutionary act that the speaker should
perform is that he must make himself physically available and mentally
ready for whatever he is inviting the speaker for.
The structure of an illocutionary act consists of propositional content,
illocutionary force, and contextual information. The propositional content
is contained in the proposition of the message. For example, the two sen-
tences You will jump o the car. and Jump o the car! have the same
propositional content you will jump o the car; though the former may be
seen as a predication and the latter as a command. The contextual informa-
tion is concerned with the context in which the speech act is uttered. A
context, suggested by Auramki et al. (1988), is delineated by the terms of
speaker, hearer, time, place, and possible world. The rst four terms simply
dene a situation in which a speaker utters something to a hearer at a
certain time and a certain place. The possible world refers to the residual
features of the context which are relevant to the performance of a speech
act. The possible world is something more than the actual world, and it
enables us to talk about what could be (e.g., the future). Terms that indi-
cate the jurisdiction (i.e. authority of the speaker over the hearer), presup-
positions of the agents involved in the performance of the speech act, and
the like are typical constituents of the specication of the possible world.
An illocutionary force is a necessary component of every successful
speech act. There are seven illocutionary forces identied by Searle and
Vanderveken (1985). They are (1) illocutionary point, (2) degree of strength
of the illocutionary point, (3) mode of achievement, (4) propositional
content conditions, (5) preparatory conditions, (6) sincerity conditions, and
(7) degree of strength of the sincerity conditions.
The most important component of the illocutionary force is the illocu-
tionary point, while the rest serve as the further specication and conse-
quences of the illocutionary point. The illocutionary point tells the point or
purpose which is internal to its being an act of that type. It determines the
direction of t for the propositional content and the commitments created;
and the direction of t determines the direction of matching between the
propositional content and the world (Austin 1980). Searle and
Vanderveken (1985) propose the ve types of illocutionary points.
Assertive This is to state how the world is (e.g., to claim, to inform, to
predict).
Commissive This point makes the speaker committed to doing something
(e.g., to promise, to oer, to accept, to refuse).
Directive This concerns making the hearer do things (e.g., to request, to
order, to permit).
Declarative This is used when the speaker wishes to change the world
through the utterance (e.g., to declare, to appoint, to approve).
Expressive This point is employed by the speaker in expressing feelings
and attitudes (e.g., to apologise, to thank, to congratulate).
Degree of strength of the illocutionary point: dierent illocutionary acts
often achieve the same illocutionary point with dierent degrees of
strength.
The propositional content conditions determine if a speech act is non-self-
contradictory and feasible. For example, a sentence I order you to have
nished your homework last week is self-contradictory; Cut a square
round shape is infeasible; that one promises to have done something in the
past is both self-contradictory and infeasible.
The preparatory conditions specify the states of aairs the speaker must
presuppose to exist in the world if an intended illocutionary act is to be per-
formed. For example, to let the hearer do something, the speaker must
assume that the hearer is capable of that. This kind of preparatory condi-
tion is normally an ontological constraint by which one can distinguish the
speech act that may be successful and non-defective from those that are
impossible.
A speech act may be successfully performed but defectively so. For
example, one may apologise to a hearer for something that one considers
disgraceful but the hearer thinks otherwise. Similarly, to issue a command
can be successfully done by the speaker; but if the performance is not
within the appropriate jurisdiction (e.g., the speaker has no necessary
authority), the speech act is defective. Therefore, if one could consider that
the success of a speech act is largely dependent on the speaker, the eect of
it relies, on the other side, fully on the hearer. In other words, the eect of
the speech act is realised through self-adjustment by the hearer, or through
the perlocutionary act.
7.2.2 Functional approach
The functional view of natural language (typied by Dik (1979, 1989))
regards a language as an instrument which human beings use intentionally
to achieve certain goals and purposes. To achieve the goals and purposes,
human beings perform linguistic actions following complex, socially estab-
lished patterns. A speaker uses linguistic expressions to communicate mes-
sages to an addressee so as to construct or modify the state of aairs in a
certain way.
According to functional grammar theory (Dik 1989), any natural lan-
guage text can be divided into clauses and extra-clausal constituents. The
clauses follow certain underlying structures and can be mapped onto the
actual form of the corresponding linguistic expressions by a system of
expression rules. The expression rules determine the form of each part of
the expression, their order, and the intonation.
The underlying clause structure is a complex abstract structure that can
be analysed at several levels:
clausespeech act,
propositionpossible fact,
predictionstate of aairs,
predicateproperty/relation,
termentity.
The constructions at the lower level are the building blocks of the higher
level. A termis an expression denoting some entity in some (mental) world.
Predicates designate types of entities and patterns of actions. Apredication
consists in a few predicates to specify relations between entities and/or
actions. Aproposition is composed of one or more predications; it connotes
a possible state of aairs. Aclause (or message) is constituted of a proposi-
tion associated with a certain propositional attitude; it normally includes
entries of speaker, addressee, speech time, content, etc. (Weigand 1990).
Predicates are organised in predicate-frames, the structure that species
the fundamental semantic and syntactic properties, such as (i) the syntactic
category of the predicate (verbal, nominal, adjectival), (ii) the number of
arguments, (iii) the semantic functions of the arguments (agent, goal, recip-
ient, etc.). For example, a predicate-frame of give can be specied as
follows:
2
give
V
(x
1
)
Ag
(x
2
)
Go
(x
3
)
Re
which states the category of give is V; it has three arguments which play
the roles of agent, goal and recipient. Predicate-frames can be extended by
satellites (non-arguments) which pertain to additional features such as
2
This and the next three examples are based on Groot (1992) and Dik (1989), of which the
former is also the source of the syntax used in these examples.
location, speakers attitude, or character of the speech act. Terms are the
NPs (noun phrases) to replace the arguments in the predicate-frames and
satellites. There are two types of terms to be distinguished: (i) basic terms,
which are expressions given as such in the lexicon (e.g. pronouns, proper
nouns, question words) and (ii) derived terms (a derived term is a basic term
with optional set of restrictors). Nuclear predications consist of predicates
and terms. A nuclear predication as a whole designates a set of States of
Aairs (SoA) which is used in the broad sense of conception of something
which can be the case in some world (Groot 1992). Core predications
are obtained by adding predicate operators and satellites. The predicate
operators relate aspectuality information (e.g. perfective/imperfective,
quanticational and phrasal descriptions) to the predications; the predicate
satellites are terms which specify manner, speed, instrument and the like for
the predications. Extended predications are further constructed with predi-
cation operators and satellites. The most important types of predication
operators are tense, external phrasal aspect, quanticational aspect, objec-
tive mood, and polarity (positive/negative) operators. The satellites are, for
example, terms with functions of time and place. Below is an example of
extended predication:
Past read
V
(John)
Ag
(the book)
Go
(quickly)
Manner
(yesterday)
Time
which reads as John read the book quickly yesterday. Notice that the
sequence of the words is supposed not signicant either in the sentence or
in the predicate-frame, though in an actual use of language the word
sequence may have some pragmatic eects. Propositions can be built from
extended predications by incorporating proposition operators and satellites.
Proposition operators are employed mainly to distinguish two modalities,
subjective and evidential; proposition satellites indicate the attitudes of the
speaker. The result of these extensions is called an extended proposition, see
below for example:
Subjunctive Mood [John writes the book] (in my opinion)
Attitude
which represents In my opinion, he should write the book. Clauses
(equally, messages or speech acts) can be composed on the basis of the
extended propositions by associating illocutionary operators and satellites.
Typical examples of the operators are declarative, imperative, and interrog-
ative; and the satellites are terms describing the action of speaking, namely
the illocutionary act itself. Below there is an example of a speech act:
Decl [he is stupid] (frankly)
Illocution
which reads Frankly, he is stupid.
The levels of term, predicate and predication are concerned with represen-
tation which is at semantic level; while the proposition and clause levels deal
with interpersonal relationships (in other words, the issues at pragmatic
level).
7.2.3 Deontic logic for communication
A communication act may convey the speakers intention which is corre-
lated to the illocution of the act. The ultimate purpose of performing a
communication act, in business cases, is to establish or alter social relation-
ships between the agents. This capacity of producing social eects is called
the deontic aspect of the communication act, which can be systematically
studied by means of deontic logic.
Deontic logic
The word deontic is derived from a Greek word which means as it should
be or duly. Deontic Logic (DL) is the study of those sentences in which
only logical words and normative expressions occur essentially (Follesdal &
Hilpinen 1970). Its subject matter is a variety of normative concepts, such
as obligation (prescription), prohibition (interdiction), permission and
commitment. The rst concept of these is often expressed by such words as
shall, ought and must. The second concept is expressed by shall not,
ought not and must not. The third one is associated with may. The last
notion is related to an idea of conditional obligation, expressible by if . . .,
then it shall (ought, must) be the case that . . ..
Standard DL
Von Wright has proposed an Old System or OS (Wright 1951) which has
had an enormous inuence on later work. It consists of the following
axioms and rules:
(C0) All (or enough) tautologies of propositional calculus
(C1) OpP p
(C2) PpPp
(C3) P(pq)PpPq
(C4) pqPpPq
(C5) O(p p) and P(pp) are not valid.
Later, it was realised that the system OS is very close to a normal modal
logic. A von-Wright-type system of a normal modal logic is presented as a
standard deontic logic OK (qvist 1984):
(A0) All (or enough) tautologies of propositional calculus
(A1) PpOp Permission is the dual of obligation
(A2) O(pq)(OpOq) If p implying q is obligatory, then p being
obligatory implies that q is obligatory
Rules of proof:
(R1) p, p qq (Modus ponens, detachment)
(R2) pp (O-necessitation)
The system OK
is obtained by adding an axiom:

(A3) OpPp Whatever is obligatory is also permitted
Furthermore, OK (and hence any normal monadic logic) is closed under
the following rules of proof, which means that every OK-theorem can be
derived from the axioms and applications of the rules of proof:
(OKa) pqpOq
(OKb) pqpOq
(OKc) pqpPq
(OKd) pqpPq
Finally, OK
is closed under the following rule of proof:

(OK
) pp
whereas OK fails to be closed under that rule.
Dyadic DL
Meyer and Wieringa (1993) summarise the system NS proposed by von
Wright to cope with conditional obligations, since these gave rise to severe
problems in the standard DL approach. In the system NS the syntax is aug-
mented by the construct O(p/q), meaning that p is obligatory under condi-
tion q. The system consists of the following axioms and rules:
(NS0) All (or enough) tautologies of propositional calculus
(NS1) O(pq/r)O(p/r)O(q/r)
(NS2) O(p/qr)O(p/q)O(p/r)
(NS3) (O(p/q)O( p/q))
(NS4) P(p/q) ( p/q)
(NS5) PpO(p/qq))
(NS6) Modus ponens
(NS7) Substitution rules
Dynamic logic approach to deontic logic
One approach is to develop deontic logic as a variant of dynamic logic rst
developed by Meyer so as to avoid some paradoxes (Meyer 1988, Wieringa
et al. 1989). This approach makes a distinction between actions (practi-
tions) and propositions (assertions). It also uses a special violation atom V
(a sanction), indicating that in the state of concern a violation of the
deontic constraints has been committed. The central notion of the
approach is represented by a modal operator [] associated with an action
. The expression []is read as the performance (execution) of the action
leads necessarily to a state (possible world) in which holds; [] means
the weakest precondition that is required to ensure that afterwards. []
will be interpreted as a modal operator of the necessity kind in a Kripke-
structure induced by the performance of actions. In this approach, the
expressions that is forbidden (F), permitted (P) and obligatory (O) are
reduced to dynamic expressions as follows:
F
def
[]V A forbidden action leads necessarily to a
sanction the notation [] stands for
necessarily , V for sanction
P
def
Fa (
def
V ) A permitted action is a non-forbidden
action, or an action that possibly does
not lead to a sanction the notation
stands for possibly
Oa
def
F() (
def
[]V ) An obligatory action is forbidden not to
do, or not performing the action leads
necessarily to a sanction
For the reduction of the obligation operator O, the negation of an action ,
denoted by , expresses the non-performance of this action.
Formalisation and translation
A system of formal deontic logic can be supplemented with denitions of
locutions in ordinary English. The formal language can be conceived of as
a structure (qvist 1984)
L Bas, LogCon, Aux, Sent
Where
(i) Bas (the set of basic sentences of L) is a denumerable set Prop of
proposition letters p, q, r, p1, p2, . . . ,
(ii) LogCon (the set of primitive logical connectives or constants of L)
is the set {, , , , , , , O, P },
(iii) Aux (the set of auxiliary symbols of L) is the set consisting of the
left parenthesis and the right parenthesis (thus, Aux{(, )}),
(iv) Sent (the set of all well formed sentences of L) is the smallest set S
such that
(a) every proposition letter in Prop is in S,
(b) and are in S,
(c) if p is in S, then so are p, Op and Pp,
(d) if p, q are in S, then so are (pq), (pq), (pq) and (pq).
Denition:
Fp
def
Pp (alternatively: Op)
Intended English readings of the connectives are shown in the following
list:
: not (more fully: it is not the case that)
: and (more fully: both . . . and . . .)
: or (more fully: either . . . or . . .)
: if . . ., then. . .
: if and only if (alternative: if and only if . . ., then . . .)
O: it is obligatory that (alternatively: it must be that . . .)
P: it is permitted that (alternatively: it may be that . . .)
The symbols , , , , and are for the ve best-known so-called truth-
functions of classical propositional logic, viz. negation, conjunction, dis-
junction, material implication and material equivalence, respectively. O and
P are to symbolise the normative (or deontic) notions of obligation and
permission. There are two other symbols: (verum in Latin) represents
some arbitrary but xed logical truth or tautology, and (falsum in Latin)
some arbitrary but xed logical falsehood, contradiction or absurdity.
In order to translate certain English locutions into the formal language L
and assign English readings to the connectives in LogCon, the following
denitions can serve as equating rules for that purpose, applicable to any L-
sentences A and B:
(D1) It is not the case that A
def
A
(D2) Both A and B
def
(AB)
(D3) Either A or B
def
(AB)
(D4) If A then B
def
(AB)
(D5) If and only if A then B
def
(AB)
(D6) It is obligatory that A
def
OA
(D7) It is permitted that A
def
PA
Impact and applications of deontic logic
Deontic logic has traditionally been used to analyse the structure of norma-
tive law and normative reasoning in law. Legal analysis and legal automa-
tion were typical areas of application of deontic logic in its early days. Only
recently, along with those application areas, has deontic logic been applied
to computer applications such as specication of normative security poli-
cies, specication of fault-tolerant systems, automation of contracting,
specication of normative integrity constraints for databases. Wieringa and
Meyer (1991) provide a comprehensive survey of applications of deontic
logic, of which LEGOLis one of the examples discussed.
The LEGOL project (Stamper 1980) is one of the examples that adopts
the deontic approach to computer assisted applications to legal reasoning.
The project, initiated in the middle of the 1970s, was to attempt to remedy
the present lack of adequate methods for analysing and designing the infor-
mation systems. A quest for a language in which the organisational rules
(or norms) can be precisely expressed was formulated, and the language is
named LEGOL (LEGally Oriented Language). The basic form of a norm
in LEGOL has the following shape:
protasisapodosis
which contains other components and can be elaborated as
conditionDagent action
The deontic operator, D, can be one of the following: obligatory, permitted
and prohibited. A norm expressed in this structure has an intended reading
if . . . then it is obligatory/permitted/prohibited for some agent to do some
action.
In the four major categories of action norms distinguished, namely
standing orders, status norms, powers of legal action, and legislative
powers, the rst two categories are related to the right set of concepts; while
the last two categories of norms are associated with the power set of con-
cepts. Two components take important positions in formulation and appli-
cation of a norm in LEGOL, which are jurisdiction and purpose.
Jurisdiction is dened as the giver and the receiver of the norm (direct
agents involved, or stakeholders). The purpose is the intentional result that
the application of the norm brings in. The purpose can be characterised by
sorts of actions in relation to some event or state in a social world, for
example, to start, to nish, to prevent or to sustain an event.
Later on, the LEGOL project has evolved into a larger scale project,
MEASUR. The language LEGOL has been developed as a general norm
denition and manipulation language, not restricted to legal modelling
only. A further discussion on LEGOL will be found in Chapter 10.
7.3 Pragmatic aspect of human communication
Communication is normally a means rather than an end. The means of
communication are employed to create, modify, and discharge social com-
mitments. An agent can be in any state as far as a social commitment is con-
cerned, for example, obliged, permitted or forbidden to full it. These kinds
of states are associated with the deontic concepts, thus they are called the
deontic states.
The minimal unit of human communication is the performance of
certain kinds of communication acts language acts. This is the basic
assumption of Speech Act Theory and the functional school. A complete
communication act can be dened as a structure consisting of three compo-
nents: performer, addressee, and message (Dietz 1992) (the message is
equivalent to a clause, or a speech act, in the functional grammar). The
message can be further distinguished as two parts: the function and the
content.
The content part of a communication act manifests the meaning of the
message as it is expressed in the proposition. The meaning (or semantics) of
a proposition is fully dependent on the environment in which the proposi-
tion is uttered. The interpretation is realised by relating the proposition to
the referent which is a universal social construct or a pattern of human
behaviour, which is performed by both agents: the speaker and the hearer.
Sometimes a propositions semantics can be obtained by referring to
another proposition in the case of indirect speech.
The function part of a communication act species the illocution which
reects the intention of the speaker. The illocutions can be grouped into
three dimensions. In one dimension, there are distinctions between descrip-
tive and prescriptive inventions, whereas, in another, there are denotative
and aective modes. Moreover, the illocutions can be associated with
dierent times, namely, now, the future or the past, as they are organised in
Figure 7.2. Each cell of the classication framework has been assigned a
title and a few representative verbs of that type of illocutions are placed in
each cell. The functions of communication are realised through use of
signs. If the illocution in a communication act is related to expressing the
personal modal state (e.g. feeling and judgement) mood, then it is called
aective, otherwise denotative. If the illocution used in a communication
act has an inventive or instructive eect, then it is called prescriptive, other-
wise descriptive. The classication on the time axis is based on the social
eects produced by the communication act: whether the eect is on the
future or the present/past.
The ultimate consequence of the communication act lies in the social
context which the speaker and the hearer both inhabit. At the social level of
the semiotic framework issues such as beliefs, expectations, commitments,
contracts, etc., are typical examples of social consequences. Table 7.1 shows
a preliminary analysis of the types of illocutionary acts and consequences
in relation to the propositional attitudes of the relevant parties before and
after the acts are performed. The labels in the rst column indicate types of
illocutions, e.g. PDf means prescriptive, denotative, future; DAp means
descriptive, aective, present/past as illustrated in Figure 7.2. The symbols
x and ~x appearing in the table represent a proposition or an event and
their negations. The illocutionary acts can be viewed as linguistic operators
that have certain social and pragmatic functions. The analysis in the table
reects an attempt to articulate their functions. A further attempt to adopt
these linguistic operators in studying semantic database operations can be
found in Chapter 9.
The above analysis, with much simplication over communications, is
based on a supposition that the communication is performed in an honest,
sincere and reliable manner. In such cases the speakers intention will be
7.3 Pragmatic aspect of human communication 95
i
n
v
e
n
t
i
o
n
mode
time
future
present/past
prescription
description
1. Proposal:
request, command;
promise guarantee
2. Inducement:
threaten, warn,
tempt
3. Forecast:
predict, assume
4. Wish:
wish, hope
desire
5. Palinode:
retract, annul,
revoke
6. Contrition:
regret, apologise
7. Assertion:
assert, report,
notify
8. Valuation:
judge, complain,
accuse
Denotative Affective
Figure 7.2. Classication framework of illocutions (based on Stamper, (1973)).
expressed in the speech act by the illocutions. As a result, there will be a per-
locutionary eect on the addressee. The analysis shown in the table distin-
guishes the addressee from the hearer. The former is considered as the
target of the speech act while the latter is someone overhearing the conver-
sation; very often they may all be relevant in communication analysis. The
social consequence is the further expansion of the eect produced by the
utterance of the act. The social consequence is an interpersonally, jointly
established relationship between the speaker and the hearer. The conse-
quences of illocutions in general are on both the speaker and the hearer,
though in some cases one may be more strongly aected than the other. For
example, in the case of the speaker making a proposal, it can be directed to
someone else or to the speaker himself. In the former case, it may be a
request or a command, and the primary eect of the speech act will be on
the addressee, which totally diers from the latter case. The social conse-
quence, produced by the social and cultural norms, will serve as a kind of
terms of reference of socially contracted behaviour for both the speaker
and the hearer.
Some explanations must be oered on Table 7.1.
(1) There are always consequences to both parties of the communication
act. The speaker and addressee each and both have propositional atti-
tudes before and after an illocutionary act. The ones before are condi-
tions, whereas the ones after are consequences. The ones in the table are
views from the speakers perspective.
(2) Directives (not being listed in the table as a basic category) are complex
acts that are composed of proposals and inducements. When a speaker
with a proper authority makes a proposal, the inducement is implied,
therefore the proposal he makes is in the nature of a directive. Without
authority, one can also produce a directive by associating the proposal
with an inducement (e.g. threat or reward). It should be noticed that the
assumed conditions of honest, sincere and non-defective communica-
tions are essential. This implies the communication will always be
carried out normally, not taking into account the abnormal disruption.
(3) Some illocutions have stronger illocutionary forces than others which
are not taken into account in the classication. For example, the illocu-
tions command, insist, request, ask and beg, all create obliga-
tions on the addressee; but command and insist have stronger
illocutionary force than the rest, and of these the former is even associ-
ated with authority. Beg has the weakest illocutionary force, but it still
creates obligation on the addressee because of the ideal conditions
assumed in the communications.
T
a
b
l
e

7
.
1
.
P
e
r
l
o
c
u
t
i
o
n
a
r
y

a
c
t
s

a
n
d

p
r
o
p
o
s
i
t
i
o
n
a
l

a
t
t
i
t
u
d
e
s
.
S
p
e
a
k
e
r

(
S
)
s
I
l
l
o
c
u
t
i
o
n
a
r
y
A
d
d
r
e
s
s
e
e

(
A
)
s
I
l
l
o
c
u
t
i
o
n
p
r
o
p
o
s
i
t
i
o
n
a
l

a
t
t
i
t
u
d
e
s
a
c
t
s

(

=
p
r
o
p
o
s
i
t
i
o
n
a
l

a
t
t
i
t
u
d
e
s
L
i
s
t
e
n
e
r

(
L
)
s
t
y
p
e
B
e
f
o
r
e
A
f
t
e
r
o
p
e
r
a
t
i
o
n
s
)
B
e
f
o
r
e
A
f
t
e
r
p
r
o
p
o
s
i
t
i
o
n
a
l

a
t
t
i
t
u
d
e
s
(
1
)

P
D
f
w
i
s
h
e
x
p
e
c
t
a
t
i
o
n
p
r
o
p
o
s
a
l
s
l
i
b
e
r
t
y
o
b
l
i
g
a
t
i
o
n
(
t
o

o
t
h
e
r
s
)
l
i
b
e
r
t
y
o
b
l
i
g
a
t
i
o
n
p
r
o
p
o
s
a
l
s
w
i
s
h
e
x
p
e
c
t
a
t
i
o
n
(
t
o

s
e
l
f
)
(
2
)

P
A
f
w
i
s
h
e
x
p
e
c
t
a
t
i
o
n
i
n
d
u
c
e
m
e
n
t
s
e
x
p
e
c
t
a
t
i
o
n
(
3
)

D
D
f
e
x
p
e
c
t
a
t
i
o
n
e
x
p
e
c
t
a
t
i
o
n
f
o
r
e
c
a
s
t
s
e
x
p
e
c
t
a
t
i
o
n
(
4
)

D
A
f
w
i
s
h
w
i
s
h
w
i
s
h
e
s
b
e
l
i
e
f
t
h
a
t

S

w
i
s
h
e
s

x
(
5
)

P
D
p
b
e
l
i
e
f
i
n

x
b
e
l
i
e
f
i
n

~
x
p
a
l
i
n
o
d
e
s
b
e
l
i
e
f
i
n

~
x
t
h
e
n
b
e
l
i
e
f
i
n

~
x
(
6
)

P
A
p
w
i
s
h

~
x
a
p
p
r
o
v
a
l

o
f
~
x
c
o
n
t
r
i
t
i
o
n
w
i
s
h

~
x
a
p
p
r
o
v
a
l

o
f
~
x
(
7
)

D
D
p
b
e
l
i
e
f
b
e
l
i
e
f
a
s
s
e
r
t
i
o
n
s
b
e
l
i
e
f
(
s
t
r
o
n
g
e
r
)

b
e
l
i
e
f
(
8
)

D
A
p
j
u
d
g
e
m
e
n
t
j
u
d
g
e
m
e
n
t
v
a
l
u
a
t
i
o
n
s
j
u
d
g
e
m
e
n
t
j
u
d
g
e
m
e
n
t

t
o

S
s
I
l
l
o
c
u
t
i
o
n
a
r
y

a
c
t
s

a
l
s
o
i
n
u
e
n
c
e

p
r
o
p
o
s
i
t
i
o
n
a
l
a
t
t
i
t
u
d
e
s

o
f
l
i
s
t
e
n
e
r
s

o
t
h
e
r
t
h
a
n

a
d
d
r
e
s
s
e
e
s
,
e
.
g
.
a
l
i
s
t
e
n
e
r

m
a
y

h
a
v
e

e
x
p
e
c
t
a
t
i
o
n
s
i
n

t
h
e

c
a
s
e
s

o
f
p
r
o
p
o
s
a
l
s
a
n
d

f
o
r
e
c
a
s
t
s
;
b
e
l
i
e
f
s

i
n

t
h
e

c
a
s
e
o
f
p
a
l
i
n
o
d
e
s

a
n
d

a
s
s
e
r
t
i
o
n
s
.
Propositional attitudes can be roughly indicated by two scales: the denota-

tive and the aective (see Figure 7.3). A propositional attitude can nor-
mally be expressed as a position on one of the two scales. The denotative
attitudes are related to someones acceptance of a proposition or event in
terms of belief. One can have a strong or weak belief or disbelief, or a
neutral position. A supposition is related to a neutral position of belief: the
neutral belief may be the condition before and the consequence after the
hypothesis is made; very positive or negative beliefs are, for example, related
to assertions. The aective scale facilitates ones indicating value judgement
such as like and dislike, and aspiring feelings such as desire and fear. Desire,
wish, hope and fear are examples of such feelings. Deontic consequences
are produced on the basis of mixtures of these two kinds of propositional
attitudes.
7.4 The Norm Analysis method
7.4.1 The concept of norms
The idea of a norm corresponds at the social level to the idea of an
aordance at the individual level. Wright (1963) explains the concept of a
norm in this way:
Norm has several partial synonyms which are good English. Pattern, standard,
type are such words. So are regulation, rule, and law.
Norms are developed through practical experiences of agents in a society,
and in turn have functions of directing, coordinating and controlling
actions within the society. Complex agents can be formed for certain pur-
poses, for example, cultural clubs, political parties, corporate bodies, gov-
ernments, nations and multi-national alliances. The functions of norms,
Denotative scale
belief
(supposition)
disbelief
Affective scale
like
desire
(indifference)
dislike
fear
+
+
Figure 7.3. Propositional attitude scales.

such as laws and regulations of the organised complex agents, are to deter-
mine of patterns of behaviour whether they are legal or acceptable within
the social context. Norms also have directive and prescriptive functions,
which are sometimes termed normatives. When a particular agent is to act,
the norms provide guidance for his action. However, in each particular
case, the agent will evaluate the situation he is in and nd some norms
which he thinks relevant to his case. The process of this evaluation and
selection of norms may be entirely subjective. In other words, norms do not
have predetermined relationship to particular actions, because the norms
can be manipulated, applied and disregarded accordingly. Holy and
Stuchlik (1983) explain the non-deterministic relationship between norms
and particular actions as follows.
(1) A norm has to do with a type of situation, not with one particular situ-
ation.
(2) Any action is understandable to at least someone other than the actor,
which means that it is performed according to some plan, rule or norm.
An action is considered as deviant or norm-breaking, but it is not
deviant essentially but only in relation to one particular norm or a
cluster of norms; it is, at the same time, performed according to some
other norms.
(3) An actor may invoke dierent norms for deciding on his actions. On
whether the action is norm-conrming or norm-breaking, one has to
speak from the viewpoint of particular norms.
Nevertheless, norms are developed as collective aordances of the complex
agent at social level. The society allows itself or its members to conduct
actions on an agreed moral or legal basis. They serve as a standard for the
members to coordinate their actions, and this makes it possible for one to
expect, predict and collaborate with others in performing coordinated
actions. When modelling the agent and the actions, which may reveal the
repertoire of available behaviour of the agent, the norms will supply ratio-
nale for actions.
There are six components of norms (Wright 1963), as follows.
(1) The character: the eect of the norm, typically, ought to for manda-
tory norm, may for permissive norm, and must not for prohibitive
norm.
(2) The content: the action or activity prescribed in norms.
(3) The condition: the circumstance or the state of aairs in which the norm
should be applied.
(4) The authority: the agent who gives or issues the norm.
(5) The subject(s): the agent(s) who can apply the norm.
(6) The occasion: location (space) or span (time) in which the norm is
given.
Von Wright calls the character, the content and the condition the norm-
kernel. Behind all these components, there are promulgation and sanc-
tion as means of enforcement for application of norms. Brkic (1970)
describes such means as systems of social control at multiple levels.
Societies employ various systems, such as normative, religio-ideological,
educationalculturalscientic, socio-political, and economic ones, as the
primary control. Other systems of organised forces in the form of police
can be used as an institutionalised form of coercion, whereas military
aggression and blackmail would be non-institutionalised forms of coer-
cion. Other lower-level instruments of control, permitted by the higher-
level systems, are, for example, a companys economic means, business
organisation policies, employees codes, etc. All these norms and means of
control form a kind of vital force which, formally or informally, explicitly
or implicitly, governs all members of the social or cultural community, or
business institution. Within the scope of this kind of vital force, people use
signs and language to establish commitments, discharge responsibilities,
full social obligations.
7.4.2 Norms in business organisations
Norms exist in a community and will govern the members of the commu-
nity in their behaviour, thinking, making judgements and perceiving the
world. The shared norms are what dened a culture or sub-culture. A sub-
culture may be a team who know how to work eectively together, and their
norms include a solution to their organisational problems. Norms can be
represented in all kinds of signs, whether in documents, oral communica-
tion or behaviour, in order to preserve, to spread and to follow. However,
one cannot always put hands conveniently on a norm, as one might grasp a
document that carries information through an organisation. A norm is
more like a eld of force that makes the members of the community tend to
behave or think in a certain way (Stamper et al. 1997).
Norms are developed through the practical experiences of people in a
culture, and in turn have functions of directing, coordinating and control-
ling actions within the culture. A research group or a working team may
have a sub-culture and therefore may have local norms. The norms will
provide guidance for members to determine whether certain patterns of
behaviour are legal or acceptable within the given context. An individual
member in the community, having learned the norms, will be able to use the
knowledge to guide his or her actions, though he or she may decide to take
either a norm-conforming or a norm-breaking action. When the norms of
an organisation are learned, it will be possible for one to expect and predict
behaviour, and hence to collaborate with others in performing coordinated
actions. Once the norms are understood, captured and represented in, for
example, the form of deontic logic, this will serve as a basis for program-
ming intelligent agents to perform many regular activities.
One way to categorise the norms is according to how norms control
human behaviour. Five types of norms can be identied in this way, each of
which governs a certain aspect of human behaviour. Perceptual norms deal
with how people receive signals from the environment via their senses
through media such as light, sound and taste. Cognitive norms enable one to
incorporate the beliefs and knowledge of a culture, to interpret what is per-
ceived, and to gain an understanding based on existing knowledge.
Evaluative norms help explain why people have certain beliefs, values and
objectives. Behavioural norms govern peoples behaviour within regular pat-
terns. Finally, denotative norms direct the choices of signs for signifying;
such choices are culture-dependent, e.g. the choice of a colour to signify
happiness or sadness.
Another method of classication of norms is based on the eects of the
execution of norms. Four types of norms are found in this classication.
Standing orders The execution of a standing order may result in a change
in the physical world. Standing orders are commands to perform
actions, and are usually expressed as one may, may not, must and
must not do something. Typically, in a business environment, a
department manager can give a standing order to the people in the
department.
Status norms The execution of a status norm often creates or alters social
structure and legal relations. The norm designates one to have liability,
immunity, right, or no-right over certain events or actions. Job descrip-
tions and company regulations are examples of these norms.
Powers of intervention The execution invokes or inhibits the use of exist-
ing standing orders and status norms. For example, a judge may have
the power to decide which law to apply to a case.
Legislative powers These change the other norms. The execution of this
type of norms may lead to an alteration of the legislative structure.
The agent applying the norm must have the right to modify or create a
norm. Normally the board of directors has the powers to alter the
company rules and to set up new rules.
Norms can also be classied according to the types of objects that the
norms are applied to. Three classes of norms are found.
Substantive norms These norms are concerned with core business func-
tions and operations. These business functions and operations con-
tribute directly to organisational objectives.
Communication norms These norms specify patterns, structure and pro-
cedures of communication within an organisation. The role of the
norms is to ensure that everyone knows what information should be
kept and provided, and what actions should be taken and when.
Without these norms, the whole organisation cannot act in a coordi-
nated manner to achieve its objectives.
Control norms These norms introduce sanctions and rewards. They act as
a mechanism to reinforce that everyone does what he or she is sup-
posed to do, as prescribed by the other two types of norms.
These classications do not exclude one another. For example, a standing
order can be any of the three types: substantive if it is concerned with
business operations; communication if it species the processes for how
the task should be performed; and control if it informs about the conse-
quence when one fails to complete the task.
7.4.3 Norm Analysis
Norm Analysis is useful for studying an organisation from the perspective
of agents behaviour which is governed by norms. From this perspective, to
specify an organisation can be done by specifying norms (Stamper 1992).
Norms can be expressed in a natural language and a formal, machine-exe-
cutable language, and they will be incorporated into the database as
integrity and consistency constraints.
A Norm Analysis is normally carried out on the basis of the result of the
Semantic Analysis. The semantic model delineates the area of concern of
an organisation. The patterns of behaviour specied in the semantic model
are part of the fundamental norms that retain the ontologically determined
relationships between agents and actions without imposing any further
constraints. A pattern of behaviour is available as long as the antecedent
behaviour is there as described in the ontology model. However, a realisa-
tion (i.e. obtaining an instance) of the antecedent may not necessarily lead
to a realisation of the dependent behaviour; there are norms to allow the
agent to judge the situation and to decide if a particular action will be
taken. Results of Norm Analysis can be written in NORMA, and further in
any programming language.
Normally a complete Norm Analysis can be performed in four steps.
1 Responsibility analysis
This analysis enables one to identify and assign responsible agents to each
action. The analysis focuses on the types of agents and types of actions. In
other words, it would answer the question as to which agent is responsible
for what type of actions.
As an instance of behaviour, its lifecycle goes from being realised,
through being sustained, till termination. The rst and the last states are
marked as the start and the nish, and in between them the behaviour
exists. One should be interested in the agents who bring about changes of
behaviour, namely, the agents authorising the start and the nish. This
analysis is to dene these two kinds of agents who inuence the start and
the nish, as they are called the start authority and the nish authority of
the aordance in NORMA.
In an organisation, responsibilities may be determined by the organisa-
tional constitution or by common agreements in the organisation. In the
CRIS case (see Appendix B), for example, to decide if a topic should be
included in a working conference is the responsibility of the programme
committee. More responsibilities are dened by the IFIP organisational
rules as follows in NORMA:
working conference @, @: IFIP Council;
work @, @: Organising Committee;
meeting @, @: Programme Committee, Organising
Committee.
The above norms detail the responsible agents for the actions. The IFIP
Council decides when a working conference starts and nishes; the
Organising Committee is the authority to recognise whether or not some-
thing is a piece of scientic work (in the IFIP context); the Programme
Committee and the Organising Committee can call a meeting.
2 Proto-norm Analysis
Consider the basic structure of a norm:
conditionDagent action.
A norm in this form species the circumstances in which the action may,
must, or must not (as indicated by the deontic operator) be performed by the
agent. The Proto-norm Analysis helps one to identify relevant types of
information for making decisions concerning a certain type of behaviour.
After the relevant types of information are identied, they can be used as a
checklist by the responsible agent to take necessary factors into account
when a decision is to be made. The objective of this analysis is to facilitate
the human decisions without overlooking any necessary factors or types of
information.
In the CRIS case, for example, the organising committee responsible for
assigning priorities to the eligible persons for attending a working confer-
ence must know the following types of information:
name of the person to be possibly invited;
whether or not this person is a member of a working group or a technical
committee; and
whether the invitee is an author or a referee.
3 Trigger analysis
This step of the analysis is to consider the actions to be taken in relation to
the absolute and relative time. The absolute time means the calendar time,
while the relative time makes use of references to other events. In Proto-
norm Analysis there is no mention about when the types of information
should be supplied, decisions made, or actions taken. But in actual business
practices, all the activities are organised in relation to time.
The results of trigger analysis are specications of the schedule of the
actions. All the actions can be organised in dynamic sequences. By setting
up and managing triggers for the actions, the automated system can
prompt human agents to respond to the situation in time. The triggers can
be rst stated in a natural language and later translated into a computable
form. The keys to the triggers are the absolute time and relative time linked
to the action types contained in the semantic model.
In the example of the CRIS case, determination of applicants priority
must be done a certain time before the invitations for the working confer-
ence are due to be sent. Moreover the invitations for the conference must be
sent to the participants in time.
4 Detailed norm specication
In this step of analysis the contents of norms will be fully specied in two
versions, a natural language and a formal language. The purposes for this
are (1) to capture the norms as references for human decision, and (2) to
perform actions in the automated system by executing the norms in the
formal language. For example, in the CRIS case, the rules of assigning pri-
orities to applicants, written in English, can be recorded and they can be
translated into the formal language LEGOL for automatic execution.
Example Box 7.1: Norm Specication
In business, most rules and regulations fall into the category of behav-
ioural norms. These norms prescribe what people must, may, and must
not do, which are equivalent to three deontic operators is obligatory, is
permitted and is prohibited. The following format is suitable for speci-
cation of behavioural norms (the examples are from Liu & Dix (1997)):
whenever condition
if state
then agent
is deontic operator
to action.
Adopting this form, a credit card company may state norms governing
interest charges as follows:
whenever an amount of outstanding credit
if more than 25 days after posting
then the card holder
is obliged
to pay the interest;
whenever an agreement for credit card is signed
if within 14 days after commencing
is permitted
to cancel the agreement.
The rst norm says that after 25 days of posting the invoice, if there is
still an amount of outstanding credit, the card holder will have to pay
the interest. The second norm states that the card holder retains the
right of cancellation of the agreement within 14 days of commencing.
The next norm says that unless there is a special arrangement made, e.g.
with the account manager, the card holder is not allowed to spend more
than the credit limit:
7.4.4 Norms in computer systems
A constraint in an information system, according to van Griethuysen
(1982), is a prescription or prohibition of the behaviour of the Universe of
Discourse or parts thereof. Norms, controlling behaviour of the agents at
the social level, will have to be reected as constraints in the conceptual
models and incorporated into the technical information systems. The
norms function as integrity constraints that safeguard the systems from
behaving improperly.
One approach is to classify the constraints into three groups (Weigand
1993).
Analytical constraints, which dene the meaning of the objects and oper-
ations in a system. For example, a student can attend a class and a
teacher can teach.
Empirical constraints, which set the rules for the generalisation of the
domain. For example, an age acceptable to a system cannot be nega-
tive.
Deontic constraints, which specify the rules that the objects and opera-
tions must comply with. For example, a student can take as a
maximum ve subjects in a semester. The special feature of deontic
constraints is that they can be violated. A violation should not lead to
an inconsistency in the information system. Instead, the system should
Example Box 7.1 (cont.)
whenever purchasing
if no special arrangement is made
is forbidden
to exceed the credit limit.
The card holders are expected to behave according to the norms stated
in the agreements. As understood by both the customers and the credit
card company, the company may impose sanctions if a customer fails to
observe the norms. With this form of specication of norms, a computer
program can be written to execute the norms. As long as the norms are
specied, computing technologies such as active databases, object tech-
nology and articial intelligence will have dierent approaches towards
software realisation.
incorporate a mechanism to derive corresponding rules of action for
sanction and correction.
Norms can be treated as these constraints. In a computer system, these
norms are programmed and incorporated into computer operations. These
constraints ensure the system behaves reasonably and correctly. The con-
straints can also function as triggers for the computer system to perform
certain actions, such as changing the value of a record or producing a
message.
8
The social layer: modelling organisations as
information systems
As the semiotic framework shows, the issues at the social level are con-
cerned with beliefs, expectations, commitments, contracts, law and culture,
as well as business functions. If a technical information system is to serve
the organisation well, to understand the organisational functions at the
social level is crucial. Information systems analysis and design must start
with understanding and modelling the organisation rst. Many informa-
tion systems methods provide techniques to deal with business analysis and
modelling. Olle et al. (1991) summarises the situation by saying that the
methods and techniques may be characterised by their dierent perspec-
tives in modelling:
data-oriented perspective,
process-oriented perspective, and
behaviour-oriented perspective.
However, by reading that textbook, one can see that they mostly concen-
trate on capturing business functions.
The semiotic approach will take one a step further. It will stress the dis-
tinctions as well as the interdependent links between the organisation, the
business process and the IT system. The notion of human responsibility
and possibility of delegation of functions to an IT system is claried. An
organisational morphology presented in this chapter will oer a useful way
for one to understand the whole organisation and will help the modelling.
Finally it will discuss the characteristics of an eective method for informa-
tion systems modelling, which suggest necessary considerations in selection
of a method.
108
8.1 Organisations as information systems
An organisation is a social system in which people behave in an organised
manner by conforming to a certain system of norms. These norms are regu-
larities of perception, behaviour, belief and value which are exhibited as
customs, habits, patterns of behaviour and other cultural artefacts. Teams
and organisations exist by virtue of their social norms and cultures. An
organisation can be characterised as a structure of social norms, which
allows a group of people to act together in a coordinated way for certain
purposes. The common purposes may be little more than to maintain the
existence of the organisation so that it may serve as an arena for the pursuit
of diverse individual and sectional interests of its members (Huang 1998).
An organisation can be seen as an information system where agents
employ signs to perform purposeful actions. Some of the organisational
functions are of high regularity where rules can be clearly formalised as
bureaucracy. Within the formalised part of the job, part of it may be highly
repetitive and can be automated where computer-based systems can be
introduced. Therefore, the computer-based system is just part of the formal
part of the organisation which is, in turn, part of the total organisation.
Stamper (1992) names this the organisational onion (Figure 8.1).
The informal information systems
The whole organisation is rst of all regarded as an informal information
system in which the oral culture plays an important role. The organisational
culture, customs and values are reected as beliefs, habits and patterns of
8.1 Organisations as information systems 109
INFORMAL IS
a sub-culture where meanings are established,
intentions are understood, beliefs are formed
and commitments with responsibilities are
made, altered and discharged
FORMAL IS
bureaucracy where form and rule
replace meaning and intention
TECHNICAL IS
machine-based system to
automate part of the
formal system
Figure 8.1. The organisational onion.
behaviour of each individual member. A healthy organisation would
possess a cohesive culture and personal beliefs, whereas, in an unhealthy
organisation, there may be considerable conicts between the organisa-
tional and personal levels. This layer is a sub-culture where meanings are
agreed, intentions are understood, and beliefs are formed. Commitments
and responsibilities are made, altered and discharged in this context
through negotiation, discussions and, sometimes, physical actions.
The formal information system
Inside the context of an informal information system, there is the layer of
the formal information system where literate culture plays a dominant role.
Rules and bureaucratic procedures are created to explicate and replace
meanings and intentions. The rules and bureaucracy specify how the work
should be carried out and how the tasks should be performed. They are
useful when the tasks are mechanistic and repetitive. The rules and proce-
dures can help achieve high eciency. People involved in the work do not
necessarily understand the meaning of the words and numbers they handle,
or the purposes for which they are used. People function as machines that
transmit and process sign-tokens. This de-skilling is the essential step
towards automation. However, such detachment of signs from their mean-
ings in bureaucracy could be dangerous. If the members of the organisation
do not understand the meanings of the signs they process, there will be no
sound basis for them to check that what they do is what they are supposed
to do. An organisation may be detached from its members, or it can run
eciently but not eectively as far as the achievement of its objectives is
concerned (Huang 1998).
The technical information system
The technical system, mostly a computer system or an IT system, is placed
inside the formal system layer. The technical system can be programmed
according to rules. It can automate some of the functions and procedures.
The computer system can be developed to read signs, shue and re-arrange
them, store and retrieve them, and nally present them. Many routine and
repetitive tasks can be delegated to the technical system to achieve
eciency. However, to introduce a technical system into an organisation in
order to automate some part of the work, there are certain prerequisites
that have to be met.
Well-dened work processes The work process and procedures should be
clearly dened. There should be clear bureaucracy in place where the
technical system can be tted in.
110 The social layer
Clearly dened human responsibility People using the technical system
should know clearly about their responsibility. Interfaces between the
human system and technical system should be clear as to what a user is
supposed to do and what a machine is expected to provide. The
purpose of using an IT system is to support the work process, but not
to hinder or to provide excuses for not doing the job properly.
Explicitly specied rules for operations Before introducing an IT system,
there should be clear specications of rules for business operations.
These rules can then be translated into the IT system or knowledge-
based system for automation purposes.
In summary, an IT system presupposes a formal system, just as a formal
system relies on an informal system.
8.2 The notion of responsibility
The notion of responsibility is crucial in designing information systems.
This has become clear in the above discussion on the relationship between
the informal, formal and technical information systems. An understanding
of the issue of responsibility may help answer the following two questions.
How is one to decide the extent to which human responsibilities should be
delegated to an IT system? What are the roles of human users when the IT
system does most of the work? An examination of the most fundamental
legal conceptions can help us to gain such an understanding.
A legal philosopher, Wesley N. Hohfeld, recognised eight fundamental
legal conceptions in two sets (Allen & Saxon 1986). In each set, the paired
concepts are opposites in columns and correlatives in rows:
Right Set: Right Duty
No right Privilege
Power Set: Power Liability
Disability Immunity
All eight concepts are related to the terms must, should, must not and
more. Each term may involve more than one of these legal concepts. For
example, a credit card holder must pay any outstanding amount of credit
within 25 days of posting to avoid incurring interest. This implies that a
card holder has the rights to pay, and not to pay till the due day, but a liabil-
ity of paying interest will occur from the due date if the amount is not paid
(Liu & Dix 1997). As pointed out by Allen and Saxon (1986), these terms
may also result in multiple interpretations and it is only appropriate to
expect a machine to assist the human in interpretation.
8.2 The notion of responsibility 111
An IT system should be seen as an assistant to the human, who can dele-
gate some of his or her responsibilities. Only when the human user has the
right and power can he or she transfer responsibilities to a machine. It
should be noted here that what is transferred are duties or job functions
rather than liability. In the same manner the government of a country can
delegate responsibilities and functions to an embassy in a foreign country
but cannot relinquish the liability for what the ambassador has done. Of
course, this does not stop users from attempting to use software to evade
responsibility: sorry I cant help, the computer says so!
8.3 An organisational morphology
In systems analysis and design, a rst step is to understand the organisation
for which the system is developed. This includes the structure and functions
of the organisation. A traditional method for organisational analysis is to
draw an organisational diagram in a hierarchical structure. The diagram
would normally represent the ocers in dierent positions and reporting
relationships. This kind of analysis is helpful in understanding the struc-
ture, but not what an organisation does.
One approach to understanding an organisation is to focus on the organ-
isational functions (behaviour) rather than just the structure. This
approach studies the morphology of the tasks and functions of an organi-
sation. Three areas of organisational functions can be identied, and hence
three types of tasks and three types of norms: substantive, communication
and control.
The substantive area The functions in this area contribute directly to the
organisational objectives. The tasks are productivity-related. The
actions are to realise the essential changes in the physical or social
world. For example, in an educational institute, the substantive activi-
ties would be teaching and research. In an insurance company, they
would be sales of policies, receipt and processing of claims, and pay-
ments to customers.
The communication area The functions in this area are about communi-
cation. Tasks performed in this category inform relevant people about
relevant facts, work procedures, what actions are to be taken, when
and by whom. Within a company, such communications are required
to coordinate the temporal and spatial use of resources for substantive
activities. Typical examples of such activities in this area are communi-
cations by sending memoranda, announcements of meetings and
events, telephone calls and emails.
The control area The control functions aim at reinforcing the whole busi-
ness system running properly, particularly the substantive and com-
munication areas. Tasks in this area include monitoring and
evaluation of substantive and communication actions, followed by
appropriate reward and punishment imposed on the agents responsi-
ble. Within an organisation, the power of reinforcement may be infor-
mal or it may be explicitly stipulated in rules and regulations. Between
organisations, it may be generated by inter-rm agreements or con-
tracts governed by law. But the power of enforcement both within and
between organisations ultimately rests upon socially established
norms within the business sector and on the wider cultural conven-
tions. The formal control norms supplement the informal norms in
cases where they are deemed insucient to ensure that every relevant
agent fulls his or her duties properly (Huang 1998).
8.4 Modelling the organisation
The substantive, communication and control areas constitute the whole
business organisation. A model of organisational morphology (Stamper et
al. 1994) can be represented as in Figure 8.2. The gure shows that an
organisation consists of three sub-areas: the substantive area x.s, communi-
cation or message-passing area x.m, and control area x.c. Each of them can
be further divided in the same way in more detail, such as substantive
message-passing x.m.s, messages about messages x.m.m, and control of
message x.m.c. Such a division can continue until the level of detail is
sucient for the purposes of analysis.
In an organisation, all three areas are necessary. The communication
functions are required to inform people and coordinate actions. The
control functions will ensure the other areas function properly. However,
based on this organisational morphology one can see that a lean and
healthy organisation should have to consume minimal resources in com-
munication or message-passing (x.m) and control (x.c) activities. It should
direct most of its resources to building up the organisational platform for
substantive (x.s) activities. An unhealthy or badly designed organisation
will have to consume a great deal of its energy in building elaborate commu-
nication subsystems and will have to rely largely on formal control subsys-
tems. These two types of subsystems comprise large parts of typical
bureaucratic infrastructures. The more elaborate these two subsystems are,
the heavier the bureaucratic burden is on the organisation.
8.4 Modelling the organisation 113
8.5 Summary: requirements for an eective information modelling method
To model an information system is to represent, by formal means, an organ-
isation in which people use signs for business purposes. A sound modelling
method must cover the issues in semantic, pragmatic and social aspects.
Issues at the three semiotic levels are closely related; Figure 8.3 explains the
relationship. The left arrow shows that in analysis and representation, the
focus must be rst on the semantic issues. A model containing a clear
description of the organisation, which may be in terms of general patterns
of actions, states, etc., is the rst basis for further analysis. The model of
this kind can be taken as a foundation on which the intentions of actions
can be discussed. Furthermore, the rationales, limits and consequences of
the actions at the social level can be addressed. But in the case of the real-
life operations, as the right arrow indicates, the social concern determines
the intentions that the speaker needs to express, and thus decide the words
and expressions to be uttered. This is the semiotic approach upheld for
understanding information systems.
To clarify the semantic problems and represent a social organisation
clearly is regarded as the most critical and dicult task. In a practical situa-
tion it is crucial to know how the people are involved in implementing a
plan or conducting an activity by translating the words into actions
through which they understand their tasks and duties. An eective method
must provide techniques for eliciting, clarifying and representing envisaged
users knowledge and requirements. For this purpose, the Semantic
Analysis method is advocated for its power in handling the semantic prob-
lems. Semantic Analysis focuses attention of analysis on the agents and the
actions. The semantic principle for solving business problems and inter-
preting language in the conduct of practical aairs is that meaning must be
whole organisation x
substantive x.s
x.c.s
message-passing x.m
control x.c
x.c.m x.c.c
substantive
message x.m.s
message about
messages x.m.m
control of
messages x.m.c
informal formal technical
Figure 8.2. Organisational morphology.
treated as a relationship between signs and actions. The actions include the
physical kind and the uses of signs (semiological kind) for fullment of pur-
poses. The semantic model delineates the boundary of concern in the
analysis and denes the meanings of the terminology used in the model.
Questions of intentions and social consequences are addressed by means
of analysing norms, which govern peoples behaviour. The analysis of
norms is based on the result of semantic analysis. With the semantically
dened words and expressions resulting from the semantic analysis, the
intentions and the social consequences can be associated with the actions
and the uses of signs in communication. The norms acquired, in addition to
the semantic model, serve as some more detailed knowledge of the business
organisation. Furthermore, the norms can be used for reasoning and pre-
scriptive analysis of the organisation, for example, planning and behaviour
simulation.
8.5 Summary: requirements 115
social
consequences
intentions
semantics
a
n
a
l
y
s
i
s

&

r
e
p
r
e
s
e
n
t
a
t
i
o
n
p
r
a
c
t
i
c
a
l

o
p
e
r
a
t
i
o
n
Figure 8.3. Practical operation versus analysis and representation.
Part two
Applications
9
From semiotic analysis to systems design
If the objective of a system analysis is to understand and specify the users
requirements, then the next objective is to produce a design of a computer
information system. The core of a computer information system comprises
a database and other application programs. This chapter will address the
semantic aspect of the computer information systems, and discuss the rela-
tionship between semantic models and database design.
9.1 The semantic aspect of databases
Data and code in a database are meaningless until someone assigns a
meaning to them and someone is able to interpret them. The meaning of
data is rooted in social and cultural conventions and norms. The assign-
ment of meaning to the data and interpretation of the data have to follow
the same social and cultural norms. The interpretation is a complex, cre-
ative act that relies on personal knowledge and understanding of the
norms. A successful use of data and derivation of information meaningful
to a user is a key issue in databases. This issue cannot be resolved by techni-
cal means alone but requires social and organisational arrangements.
There can be two kinds of meanings distinguished: intension (sense) and
extension (or reference). These two notions can be applied to predicates,
propositions and sentences. The intension of a predicate, i.e. its sense, is
identied with the property it expresses; its extension is the generalised class
of referents in the world of aairs that possess the property. The intension
of a proposition is determined by the meanings of the predicates used; the
extension is its correspondence to the world of aairs. Then the meaning of
a sentence is determined by the propositions it contains. People sometimes
do not know the extension (reference, denotation) of the components of a
sentence, but only their intension (sense, understanding). However, they are
119
able to understand a sentence intensionally without knowing its correspon-
dence to a world of aairs (Thayse 1989). What is critical in modelling is the
intensional meaning. The aordances in NORMA that are considered as
the foci in modelling are the abilities of sense-making, understanding, or
recognising of the intensional meanings. These abilities are generalised pat-
terns of behaviour of the agents that are governed by the social norms.
The semantic database takes an ontology model as the conceptual
schema. Aordances can be roughly comparable with entities and relations,
and determiners (a specic kind of aordance) with attributes in terms of
relational databases. Aordances are dened and organised according to
the ontology model. Instances (or particulars of aordances) are called
realisations of the aordances and are managed in the databases. Each uni-
versal and particular aordance has temporal attributes, for example a start
and a nish, as its intrinsic properties. Each application in such a database
will involve both the semantic and temporal operations.
9.2 Capturing the semantic aspect
The ontology model delineates a context which involves concepts and ter-
minology used in a particular problem domain. According to van Dijk
(1981), the context is both a theoretical and a cognitive abstraction, viz.
from the actual physicalbiologicaletc. situation. The context in which a
certain speech act is performed supplies the more general linguistic and
other knowledge in memory, against which the language users match
incoming information. The context makes it possible for agents to establish
a link between the intensional and extensional meanings. The categorised
concepts and terminology according to the ontological relationships
provide determined semantics contextually, because every word or expres-
sion is linked with its antecedents in the ontology model. For example, a
student can be dened in many ways: even someone who follows an evening
course of ball-room dancing once a week can be called a student in that
context. But, with the help of an ontology model specifying exactly the cir-
cumstance, the meaning of student can be uniquely determined by onto-
logically linking the terms used in the discourse: someone registered in a
school, only during his registration period. Thus, the NORMA expressions
(university#employer, person#employee)employs,
and
(university, person#student)registered
constitute a practical circumstance in which the boundaries and semantics
of the concepts are settled by linking the dependants to the ontological
120 From semiotic analysis to systems design
antecedents. The antecedents (e.g. in this case, university and person),
moreover, are ontologically dependent on some other antecedent (e.g. a
nation). The graphic representation may explain better how a student
and a teacher are dened (see Figure 9.1). The student and employee are
role names for a person who is registered at or employed by a university;
A teacher is a specic kind of employee; registered and employs are
aordances that are ontologically dependent on the antecedents, the uni-
versity and the person.
A denition of ontological dependency in NORMA can be given as
follows. Given two objects x and y, if ys existence depends on xs, and y only
exists during or within the existence of x, then the dependency relationship
between x and y is called an ontological dependency. Or it can be said that y
ontologically depends on x. The object x is called the antecedent and y is the
dependant. The structure constrained by the kind of ontological relation-
ship determines exactly the boundaries of each concept. In the example dis-
cussed above, the student never exists beyond the existence of the person,
the university and the registration; the same principle applies to the teacher
and administrative sta.
9.3 Capturing the time aspect
Winograd and Flores (1987) state that, in communication, time is not an
incidental condition, but a critical aspect of every speech act. Therefore
they build a mechanism for keeping temporal relations in the computer-
assisted communication system, The Coordinator. Many related works
9.3 Capturing the time aspect 121
e
m
p
lo
y
e
r
e
m
p
l
o
y
e
e
t
e
a
c
h
e
r
a
d
m
.

s
t
a
f
f
s
t
u
d
e
n
t
employs
registered
university
nation
person
Figure 9.1. Ontological structure and semantics.
(e.g. Auramki et al. (1988), Weigand (1990)) suggest that, regarding the
context of a communication act, time should be one of the elements for
specifying the context. The temporal aspect is important in relation to the
development and utilisation of information systems for at least three pur-
poses: determination of boundaries of concepts that are expressed in lan-
guage; description of dynamics of a social world or an organisation;
coordination of human activities.
Time associated with the signs, rst of all, can clarify the actions and
meanings. Any use of a sign, i.e. a speech act or a physical act, brings about
a meaning determined by examining the context, of which time is one of
the elements. This is also the case if one wants to be able to equate the
morning star with the evening star. For example, someone says that it
was bad weather the day before yesterday. To understand properly what is
meant by bad weather, someone else must be referred to the time men-
tioned in the utterance; only then can he understand that the bad weather
just means it was a bit too cold, though some people may not think it was
bad weather at all.
The use of language enables agents to talk about actions, states and
events in the direct environment or at a distance. They may refer to some-
thing that happened in the past, or will happen in the future, or something
they wish to avoid. Time plays a key role in such uses of language. For
example, John informs his boss Mary that his colleague Henry is taking
leave. In this case, it is important to know both when Henry left and when
John told Mary. Things may get more complicated if John said to Mary
that Henry told him to ask Mary for a permit for his leave. Then the time
becomes essential, because appropriateness of Henrys and Johns behav-
iour is judged, to a large extent, on whether Johns informing Mary took
place well in advance of Henrys leave.
The temporal aspect of information has a great importance in coordinat-
ing actions in practical aairs. Perlocutionary eects will result in social
consequences which have to be expressed in the time horizon. For example,
an obligation is established for a certain period of time; a retraction of
some speech act is to end the existence of an existing propositional attitude.
An illocutionary act may take place at one time; the perlocutionary act may
be at another time because of communication delay or time needed for the
hearer to take a decision. Interpersonal relationships, e.g. commitment and
responsibility, are set up at one time, altered or nished at another time.
This kind of beginning and ending processes continues constantly. Every
agent in a social and organisational context is involved in a network of con-
stantly changing relationships. Commitments and responsibilities are only
sensibly clear when time is taken into account. Time can be useful for
checking the sincerity of a speaker and feasibility of actions if the speaker
promises or plans to do too many things at the same time, then either the
speakers sincerity or the feasibility of performing the actions may be ques-
tionable. The commitments and responsibilities are normally placed into a
temporal framework so that the actions (and the hearers reactions) can be
expected and coordinated.
There are other reasons for arguing that time is an important aspect of
information. As a result, the semantic data structure for database design
must contain time as an indispensable part. That means every data record
in the database comprises times, e.g., the start time and the nish time of
the realisation of an agent or an aordance.
9.4 Ontological modelling for conceptualisation
The primary purpose of Semantic Analysis is to assist an analyst and a
problem-owner to articulate the requirements by focusing on clarifying the
language used in expressing the problem. The ontology model obtained
from the Semantic Analysis, however, serves as a database conceptual
schema (a term dened in Griethuysen (1982)).
A Semantic Analysis is a process of conceptualisation of a business
organisation through which the behaviour of the organisation is analysed
and captured in the ontology model. The agents and the patterns of behav-
iour are the focus of representation in an ontology model. The semantic
primitives appearing in the model represent possible patterns of actions of
a complex agent, which are known as aordances in NORMA. The bound-
ary or existence period of an aordance in NORMA is bound to the agent
who possesses it, while, in many other information analysis approaches (e.g.
Chen (1976), Verheijen & Bekkum (1982)), entities are supposedly observ-
able independently of the agents involved.
The aordances captured in an ontology model compose a conceptual
representation according to which the data can be organised in a database.
The realisations of the aordances are the instances that are recorded in the
database records. For example, an expression in NORMA,
(university person) employs,
denes a part of the conceptual model, which can be represented as an
ontology chart (Figure 9.1). The aordances of university and person are
two antecedents that jointly aord a pattern of action employs. The par-
ticulars of universities, persons and employments will be instances to be
stored in the database. Figure 9.2 shows some instances of these three
aordances in a database.
9.4 Ontological modelling for conceptualisation 123
9.5 Intentions, propositional attitudes and consequent operations
The semantics of language used in the problem domain is handled with
NORMA in Semantic Analysis. Semiotic behaviour is expressed in the
formula: B Ax, as discussed in Chapter 6. Using signs enables the agent to
talk about actions, states and events at dierent times with intentions. A
proposition can be uttered, for example, as an assertion, a prediction or a
hypothesis, which are coined as intentions. An extension is made to the
NORMA syntax for the association of the intention with the formula of
semiotic behaviour as follows:
B Ax , where B is the agent performing the semiotic action with an
illocution , between the quotes are signs employed by B in
his behaviour. Ax is the meaning or referent of the semiotic
action. E.g. Mary John happy assert.
The intention operator can be assigned dierent illocutions for indicating
the intentions of the agent in using the signs. The illocutions by the speaker
are, as discussed in Subsection 7.2.1, categorised under eight headings.
Following the illocutions, there will be consequent actions by the speaker
and by the hearer. In the case where a database is involved as a medium of
recording and assisting the communication, there will be corresponding
operations on the database, as illustrated in Table 9.1.
The directives will bring the hearer into an action, as normally the direc-
tives produce social consequences of deontic nature. The types of deontic
consequences that are recognised in deontic logic are obligatory, permitted,
and prohibited which are typied by auxiliary verbs in speech acts such as
should, should not, may, etc. Another type of deontic consequence that
may be produced by the directives is a liberty (or freedom) (Stamper 1980).
This is comparable with Hohfelds legal concept of privilege (or immunity),
because havinga privilege is legally weaker thanhavingaright. The privilege
enables an agent to preserve discretion in taking actions. Some of the direc-
University
Staffordshire University
Twente University
Person
Susan
James
John
Jack
Employment
Susan, Staffordshire University
James, Staffordshire University
John, Twente University
Jack, Twente University
Figure 9.2. Instances of three aordances.
tives andother illocutions by the speaker generate a liberty. The deontic atti-
tudes of obligation, permission and prohibition can be created by the direc-
tives. Some of the illocutions do not involve deontic modalities at all. This
can be seen from the fact that if the hearer does not follow the expected
pattern of behaviour, there would not be a punishment enforced on the
hearer. For example, if a speaker makes an illocutionary act of forecast by
uttering Mr X will arrive here tomorrow, it is at the hearers discretion
whether to believe or not, though normally the hearer will believe what the
speaker utteredif the communicationis honest andsincere. Therefore, Table
9.1 only represents some ideal patterns that are based on the supposition
that the agents involved in communication behave normally (i.e., obey
social norms of sincere, honest communication).
This table attempts to depict a part of a systemof social norms of com-
munication, especially focusing on the consequent actions by both speech
agents. Operations on the database, which is used to record and assist the
communication, are derived as seen fromthe table. In most of the cases, the
operationonthe database is torecordthe speechact except for the illocution
of retraction. The illocution of palinode or contrition is used by the speaker
when an earlier speech act is to be withdrawn. In this case, an operation of
correction is required on the database. All operations on the database com-
prise three aspects, semantics, intention and time, which will be discussed in
more detail after the semantic template is dened in the next chapter.
9.6 Other aspects of databases: facts, beliefs, and knowledge
A database, at the syntactic level, can be viewed as a well organised, but
variable, set of data packages, which during a longer period is used in an
9.6 Other aspects of databases 125
Table 9.1. Illocutions, propositional attitudes and database operations.
Propositional attitude
Illocutionary category of addressee (A) Operation on database
by speaker (S) (after speech acts performed) (to reect Ss acts)
Proposal obligation record (or insertion)
Inducement expectation record (or insertion
Forecast expectation record (or insertion)
Wish belief record (or insertion)
Palinode retraction update
Contrition retraction update
Assertion belief record (or insertion)
Valuation judgement record (or insertion)
organisational system as a means of communication (denition of
Lindgreen (1990)). The languageaction approaches regard a database as a
model of some Universe of Discourse for communication (see Weigand
1990)); a database functions as a device for recording, processing, indexing,
and transmitting communication acts. From this point of view, what is
managed in a database is not the recording of facts, as suggested by
Nijssen and Halpin (1989), but subjective observations, opinions and
beliefs. Based on the two radical subjective axioms adopted, knowledge is
bound to the knower. To retain the subjectivity, both knowledge and the
knower will have representations in a database. Personal value judgements,
the speakers intentions, and dierences in perspectives and interests seem
no less important than the propositional contents of the utterances, which
should be accommodated in the database as well.
From another dimension of database studies, numerous researchers have
noticed that, with many databases, information in the system reects only
the current state of aairs (e.g., Cliord & Tansel (1985), Kim et al. (1990),
Sadeghi et al. (1988)). The databases of this kind only provide a snapshot of
an organisation. However, in business practice, historical information is
important. Based on his historical studies and business experience,
Achueler (1977) strongly appeals for a database technique that will be able
to keep historical records. He describes the common update technique in
the database by erasing the records as like the Aton-Update, a mass
update of a large Egyptian database pertaining to Aton that was under-
taken with hammer and chisel throughout the country in 1347BC. This
kind of erasing update is also comparable to the update of a most impor-
tant database in the year 47BC that the Library of Alexandria was com-
pletely updated by a re. Loss of historical data through such actions on
databases may sometimes cause irreversible, disastrous consequences.
Meanwhile, the historical information can be useful for businesses. In
banking, for example, information about the history of clients can help the
banker assess the customers credibility. Therefore it assists them in making
decisions about granting loans. When a company plans to introduce a new
product into the market, historical information can help the company do a
trend analysis and predict the market for the product. On the other hand,
hypotheses and predictions for the organisation are relevant for studying
organisational behaviour and performance, for coordinating future actions,
and for planning and policy-making purposes.
The semantic temporal database developed in the project aims at accom-
modating data pertaining to business semantics and intentions. Each par-
ticular instance of actions and use of signs is associated with the intention
that is expressed by the language user. Moreover, each instance has a pair of
times, the start and the nish, to indicate the existence of an action or a
state, from the speakers point of view.
Some other concepts, such as illocution and attitude, are added to the
chart in addition to the aordances identied in the early analysis for dis-
cussion of the aspects of communication acts. With such an ontology
chart, one can represent a fact such as John occupies/rents/owns Marys
house (in a certain period), and all the three sentences (S1), (S2) and (S3)
Example Box 9.1: Semantic, intentional and temporal
aspects in speech acts
To show the distinctive components of semantic, intentional and tem-
poral aspects that can be considered in the representation with the pro-
posed data structure, the following sentences will be used in discussion
(after Searle, 1983, p. 30):
(S1) John wants Marys house in the summer of 1994 (suppose now
summer of 1993).
(S2) In 1994, John will want Marys house.
(S3) John now wants Marys house in the summer of 1994 though by then he
will not want it.
A semantic analysis of the sentences reveals there are a few universal
aordances which should appear in the ontology chart: person and
house. John and Mary are instances of the aordance person. The
word want seems to be used with mixed meanings. To want may mean
to want to have; to have may be more specically understood as to
occupy (or, to live in), to rent, to buy or to own. At the same time,
to want may indicate an intentional modality. In order to represent the
semantic and intentional aspects associated with the word want,
several possible interpretations are captured in the analysis (e.g.,
occupy, rent and own). The temporal aspect of every aordance is
treated as an intrinsic property, that is, every aordance is attached with
a start time and a nish time to delineate its existence. The temporal
aspect of every aordance is intrinsically handled by referring to the
times dened in a standard time framework. Therefore the time now,
1994 and the summer of 1994 are not explicitly shown in the chart,
but handled as hidden parts in all aordances in the ontology model.
Figure 9.3 shows the ontology chart resulting from the semantic analysis
of the three sentences.
(they are either predictions, assertions or requests). The dotted lines
linking the illocution and the attitude to the aordance have (with three
specic kinds of meanings listed under it) can indicate the illocution and
the attitude attached to the utterance. As shown in the gure, time is
dened in a standard temporal reference system that can be referred to by
every aordance. There are some important issues in the chart worth
discussing.
To grasp the meanings of the word want, several possible interpreta-
tions are identied on the chart, which allows the language user to
determine the meaning of want between occupy, rent and own
when the word want is employed in the sentence.
An illocutionary act can be made with or without an illocutionary verb
explicitly in a sentence. In the latter case, however, it may require
careful analysis to decide to which illocutionary category the sentence
belongs. For example, the sentence (S1) (John wants Marys house in
the summer of 1994) may be regarded as a prediction, an assertion, or
a request.
Intentions may be dierent frompropositional attitudes. For example, if
Ii e e io
e o
ye
c le -
c e ic-
i c l -
e o
ho e
e c i e
e c i e
illoc Iio
e icI
e I
e e I
IIiI e
e i e
v I
like
h ve
occ y
e I
ov
Figure 9.3. Semantic, intentional and temporal aspects in an ontology chart.
one says I want your house it may appear to be analogous to I like
your house (Searle 1983). In addition, one may very likely encounter a
situation in which a speaker has an unstated, dierent propositional
attitude fromthe expressed intention. However, in most of the analysis
in this book, an assumption is that communication normally takes
place inasincere andhonest manner, whichmeans consistencybetween
the intentions expressedinillocutions andthe propositional attitudes.
With the chart (Figure 9.3), one can describe the present situation (the
house is presently occupied by a person, which is just one of the possible
specic forms of have) and the dierent scenarios described by the three
sentences. The situation at present can be described as in the following
expression:
(B, house)occupy the house is now occupied by the person B (the
person B may be John or Mary, for example).
In steps, the following can be obtained in analysing the sentence (S1):
(R1.1) (A, house)occupy person A occupies a house;
(R1.2) (A, house)occupy, summer of 1994 in the summer of 1994, the
person A will occupy the
house;
(R1.3) (A, house)occupy, summer of 1994 a semiotic act (i.e. a speech
act) with the contents
between the quotation
marks;
therefore, as a whole, the following is obtained:
(R1) A (A, house)occupy, summer of 1994 want
The expression (R1) represents the sentence (S1). The contents between the
quotation marks are propositional contents which are seen as the semantic
part of the whole expression the person A wants to occupy the house in the
summer of 1994. Note that the time summer of 1994 is related to the
action (A, house)occupy to express that occupation of the house will take
place in the summer of 1994.
In steps, the following can be obtained in analysing sentence (S2):
(R2.1) (A, house)occupy person A occupies a house;
(R2.2) (A, house)occupy a semiotic act (i.e. a speech act) with
the contents between the quotation
marks;
(R2.3) A (A, house)occupy want a speech act performed by the person
A with an illocution predict;
(R2.4) ((A, house)occupy want), the act of wanting will take place in
summer of 1994 the summer of 1994;
(R2.5) ((A, house)occupy want), a semiotic act (i.e. speech act) with
summer of 1994 the contents between the quotation
marks;
therefore, as a whole, the following is obtained:
(R2) A A (A, house)occupy want), summer of 1994) predict.
The expression (R2) is from a semantic analysis of the sentence S2. It is a
prediction that in the summer of 1994 the person A will want to occupy the
house.
The analysis of the sentence (R3) in steps results in the following:
(R3.1) A (A, house)occupy, same as in (R1);
summer of 1994 want
(R3.2) (A, house)occupy not not want is an illocution to indicate
want a retraction of the speech act per-
formed earlier which is the content
between the quotation marks;
(R3.3) ((A, house)occupy not the act of not want will take place in
want), summer of 1994 the summer of 1994;
(R3.4) A((A, house)occupy not the person A has made a prediction
want), summer of 1994 predict that in the summer of 1994 A will not
want the house (which is now occu-
pied by B);
therefore, a complete expression by combining (R3.1) and (R3.3) is
obtained as follows:
(R3) A (A, house)occupy, summer of 1994 want ,
(R3) A (A, house)occupy retract-want, summer of 1994 predict.
The expression (R3) results from the sentence (S3) which says that the
person A now wants to have the house in the summer of 1994 while it pre-
dicts that in the summer of 1994 he will not want the house. In the expres-
sion, the mood will not is represented by the illocution of prediction.
The three aspects of information can be seen from the above analysis of
the three sentences. The core of the meaning is conveyed by the proposition
which can be interpreted in a semantic sense without there being any inten-
tion associated with it. The meanings of the propositions are determined by
putting the words in the semantic context in the ontology chart. The inten-
tional aspect of the semiotic act is reected by the associated illocutions.
The time aspect is dealt with as a necessary component in every expression.
All these three aspects will have to be accommodated in the semantic tem-
plate, to be discussed in the next chapter. The semantic template will be
dened as a uniform syntactic structure for databases that can handle the
semantic, temporal and intentional information.
Example Box 9.2: A speech act in a database
The three aspects of information have been identied to be represented
and processed in the semantic temporal database: semantics, intentions
and time. A standard record structure based on the elements of a speech
act can be derived. Consider the following fragment of an ontology
model:
(person#absentee, course)absence
which can be represented in the form of an ontology chart (Figure 9.4).
With this ontology model, one can record in a database facts as to who is
absent from which course at what time. In addition to that, it is also pos-
sible to record in the database the more sophisticated uses of signs, such
as hypothetical or declarative uses. The two appearances of illocution
in the chart show the possible recursive uses of signs, that is, an illocu-
tionary act can be contained in another, and so on.
Suppose the following speech comprising a piece of knowledge:
Susan yesterday observed that John had been absent from 1-6-90 till
30-6-90. This piece of knowledge can be captured in a record as
follows:
Susan, John, absence, course 123, 1-6-90, 30-6-90, assert, assert,
yesterday, .
The explanations below will assist reading of this record and also
present considerations of some issues, such as agent, intention and time,
in knowledge representation in the database record form.
(1) Susan is the agent who uttered the speech act, the user of sign.
(2) The semantic part of Susans action resides in the proposition which
is abbreviated as John, absence, course 123, 1-6-90, 30-6-90.
The meaning of John, absence and course 123 is determined by
referring to the corresponding part of the ontology chart.
(3) 1-6-90, 30-6-90 refers to the period during which an instance of
Johns absence is realised.
(4) Both the start mood (i.e. mood) and the nish mood (i.e. mood)
of the proposition, found in Susans action, are assertive. This is
because what is represented is an observation, which can be seen
from the original sentence (if it is not directly seen from the speech,
it is up to the analyst to decide what mood Susan had).
(5) Yesterday indicates the start of Susans observation. The blank
nish time means the assertion is valid till now (a lled nish time
would suggest a retraction of the assertion, which would mean
Susan had withdrawn the assertion).
a
b
s
e
n
t
e
e
person
illocution
illocution
absence
course
Figure 9.4. A chart of personcourseabsence.
10
Semantic temporal databases
The method of Semantic Analysis oers a means of representing informa-
tion requirements and database modelling. Based on the database theory
and semiotic methods, a semantic temporal database is proposed. The
semantic temporal database provides a technology of management of the
data with semantic and temporal properties. A semantic temporal database
language, LEGOL, has been developed for the specication of norms.
LEGOL statements can specify constraints and triggers, and can be exe-
cuted by a LEGOL interpreter.
10.1 Databases
A database, from the information technology point of view, can be
regarded as a kind of electronic ling cabinet (Date 1995, p. 2). It can be
dened as a persistent collection of logically related data that allows shared
access by many users. A DBMS (DataBase Management System) is the
application-independent software that protects and manages databases.
Since the late 1960s when the pioneering DBMS, IMS, was published, the
database technology has provided ever improving tools for information
management. Since then, the development of new types of DBMS and
research into new ways of organising information have been taking place.
This section provides a quick review of some major contemporary types of
database systems and of the presentation of the semantic temporal data-
bases which are considered highly relevant to business information man-
agement purposes.
10.1.1 Developments in database management systems
Relational DBMS is the most widespread and popularly used type of data-
base systems nowadays. Other types of systems are also being introduced to
133
meet diversied application purposes, typically the objected-oriented data-
bases and deductive databases.
A relational database organises data based on a model published by
Codd (1970). A relational DBMS provides the following functionalities
(Loomis 1990) which later become a standard for all kinds of DBMS.
Persistence A database exists outside of the scope of any particular
program run-unit. It has a complete separation of the data from the
program that creates and uses them. The data exist on non-volatile
storage even after the programs terminate.
Secondary storage management A DBMS provides ecient ways to rep-
resent and access small and large volumes of data with the techniques
from structuring databases, addressing database storage, providing
fast paths to database contents, clustering portions of a database in
the physical devices.
Concurrency A DBMS allows more than one user to have access to the
database concurrently without jeopardising the database and the data
integrity.
Recovery This feature enables the DBMS to cope with system failures,
especially soft crashes, so that it provides a protection of database con-
tents from destruction.
Ad hoc query facility This enables a user to access database contents
without writing a program.
A relational database design is a process of rstly identifying entities and
attributes that represent application objects and then organising them into
relations (i.e. tables). The relations must conform to formalisation stan-
dards for optimisation to achieve performance eciencies. SQL has
become the de facto relational database language. Example methods for
database design can be typied by Chen (1976), and elaborate discussions
can be found in Date (1995).
The deductive, or logic-based, approach sees the database as a set of
axioms. The ground axioms, as well as the deductive rules, are treated as
tuples stored in the database. A formal denition of a deductive database in
Thayse (1989) is given as a theory composed of an extensional database
and an intensional database. The extensional database is a set of ground
instances (i.e. ground axioms) of atoms dening the extensions of the base
predicates (i.e. a relational database). The intensional database is a set of
deduction rules dening the virtual predicates. To perform a query in the
database is to prove that some specied formula is a logical consequence of
those axioms; in other words, it is to prove a theorem. This approach can be
134 Semantic temporal databases
seen from Minker (1988), Reiter (1984), Thayse (1989) and Wieringa et al.
(1989). Propositions about some domains or Universe of Discourse (UoD)
are normally the sources of knowledge that will be organised in the data-
base. Such a type of database is sometimes termed a logic database, infer-
ential database, deductive database, knowledge base, etc. Language in
this approach plays an important role in knowledge acquisition and repre-
sentation because of the great metaphorical resemblance between the data-
base modelling and natural language uses (see Way (1991) for the notion of
metaphor in knowledge representation). One of the advantages of the
deductive over the relational database is the declarative operations on the
database, both the data input and application: what needs to be specied is
concerned with the what rather than the how. However, shortcomings of
the deductive database are lack of structure of the data and the shallowness
of the semantics of the predicates.
To cope with these problems, Weigand (1990) proposes a linguistically
motivated approach to knowledge-based systems. According to his sugges-
tion, a lexicon can be built in which the vocabulary of predicates are seman-
tically dened, so that the deep semantic structures can be used by a
knowledge engineer in conceptual modelling. In some sense, the ontology
model resulting from the Semantic Analysis is actually used as a suitable
formalism for organising the deep semantic structures of the vocabulary. It
contains certain vocabulary that is relevant to an articulated problem
domain. The words reecting the meaning in human actions are organised
in accordance with the ontological dependencies between the actions. In
this way the semantics of the words are bound to the context and uniquely
dened; therefore, uses of the word in operations on the database can be
related to the established structure and will not result in ambiguity.
Another type of database system is the Object-Oriented Database
Management Systems (OODBMS). This type of DBMS covers the func-
tionalities listed for the relational DBMS. In addition, they support the
object-oriented programming language notions of complex objects, object
identity, encapsulation, types or classes, inheritance, overriding and late
binding, extensibility, and computational completeness. The incentives for
developing an OODBMS come from the application areas of CAD, CAM,
CASE, GIS, etc. where complex data types need to be handled and a rela-
tional DBMS cannot meet the requirements. Besides, OODBMS can be
integrated with an object-oriented programming language, which means a
model of representation can be implemented by a programmer consistently
in one environment, whereas in a relational database project the application
and the database parts are two separate tasks. Standards for OODBMS, or
10.1 Databases 135
a new generation of DBMS with object-oriented functions, are proposed
by Atkinson et al. (1990) and Stonebraker et al. (1990).
New research directions of DBMS include distributed database systems,
federated architecture and integrated heterogeneous databases, client-
server architecture of distributed databases, and integration of relational,
object-oriented and knowledge databases.
10.1.2 Semantic temporal databases
The appeal of temporal databases comes from various application areas,
such as CAD, CAM, software engineering, business data processing, etc. In
some areas of business data processing where legal responsibilities are
involved and time is a critically important factor in the business transac-
tions, to keep historical images of business becomes essential. Examples of
such areas are banking and insurance industries, policing, and government
administration systems such as social security. In these applications, a key
system requirement may be the ability to make non-destructive, retrospec-
tive updates. In the design of the computer system for Income Support
(now operational) of the Department of Social Security of Great Britain, a
key requirement is to make retrospective update to claimant data, without
destroying the original data (Shearer 1992).
Data are neither real actions, events, nor states of aairs, but they are
simply a representation of what the business world believed at a given point
of time. In other words, data are nothing more than assertions about busi-
ness reality, which may only temporarily remain true at a specic time.
Therefore, one can recognise two types of time: the time referring to the
events and states, and the time referring to when the assertions are made.
Both kinds of time are sometimes important and many database systems
keep records of them.
The evolution histories of entities, actions and states of aairs in the
practical world need to be kept in the database. Databases containing these
types of data with historical images are called temporal databases (Kim et
al. 1990, Snodgrass 1985).
1
Two types of times associated with each record
are kept in the temporal databases: the event time and the transaction time.
The event time indicates the period of the event, action, or state of aairs;
the transaction time reects when the operations of entering and changing
data in the databases are actually done. Both types of time are intervals that
are marked with a start and a nish.
1
Many research eorts have been put into the temporal databases, see for example Cliord &
Tansel (1985), Sadeghi et al. (1988), Kim et al. (1990).
The database developed by the MEASUR research team has the features
of managing time at dierent levels. It requires the adoption of the ontol-
ogy charting as the conceptual modelling so that the semantic richness is
captured in the databases. The databases under such a DBMS that handle
the temporal features and organise data according to the semantic ontology
model are called the semantic temporal databases. As to the two types of
time managed in the semantic temporal databases the event time and the
transaction time the former represents the existence of an aordance, the
latter the actual time of entering and changing the content of the data
record. The event time in business applications tells the state of aairs in
relation to time periods, for example the period that one is in a particular
salary rank or nancial status. This type of time is directly related to the
depiction of dynamic pictures of practical aairs. The transaction time reg-
isters when an operation on a particular data record is performed. The
users of the databases do not have control over this type of time, in the
sense that they can only view the time when the data record is put into the
database, and when an update is made. But it is not possible for the users to
change the transaction time as the time is recorded by the DBMS automat-
ically. Table 10.1 shows the uses of the two types of times and their func-
tions.
When an observation is made explicitly about business aairs, it can be
recorded in the database, for example the status of John and Paul as to
whether or not they are students. The rst observation as seen in the rst
record in the gure is about John being a student from 20/8/86, which is
entered into the database on 25/8/86. The next observation is made to tell
that Johns studentship is from 20/8/86 till 15/6/92; the time of entering this
observation is 16/6/92. At the same time, a nish transaction time 16/6/92
is put onto the earlier record about John for labelling the record as being
10.1 Databases 137
Table 10.1. Illustration of functions of time.
Valid time Transaction time
Person Status Start Finish Start Finish
John student 20/8/86 125/8/86 116/6/92
John student 20/8/86 115/6/92 116/6/92
Paul student 20/8/86 125/8/86 16/12/92
Paul lab assistant 11/5/89 115/5/89 114/1/91
Paul lab assistant 11/5/89 31/12/90 114/1/91
Paul student 20/8/86 15/12/92 16/12/92
out-of-date. To read the tuples about Paul, one can also obtain a picture of
Pauls movement: he began to be a student from 20/8/86. During his stu-
dentship, he was a lab assistant for the period from 1/5/89 till 31/12/90. The
nish of his studentship is on 15/12/92.
Several details must be noticed about the semantic temporal databases.
Each tuple records a message which may originally be from a written
document, business report or oral notice. This shows a correspon-
dence between the database tuples and semiotic acts.
Values of the transaction time may be dierent from those of the event
time if the database only records what takes place in the business
world. However, the more one relies on the database for the business
operations, the less the dierence will be between the two types of time.
For example, if one decides to use the start of the transaction time as
the start of the event time, which means the transaction time has a
legal eect on determining the period of studentship, then these two
types of time will always be the same (therefore, only one type of them
is needed).
A tuple with an unlled nish transaction reects the current state of
aairs; therefore, there is no need to physically remove the out-of-date
records to update an observation.
10.2 The semantic templates
The purpose of dening a semantic template is to devise a uniform struc-
ture by means of which the semantics and intentions of use of signs can be
accommodated and operated with. The semantic template will be the basis
for devising a uniform database structure.
10.2.1 Dening a semantic template
The semantic primitives handled in the knowledge representation language
NORMA are the words that label the aordances. The words are used to
describe practical problems in a business context with intentions. In order
to capture the semantics and intentions of the use of words, behind every
NORMA expression there is a uniform semantic structure, which is termed
the semantic template. A Semantic Template (ST) is dened as follows.
Denition of the semantic template:
ST : agent, ST, mood, mood, time, time agent,
action, time, time ;
agent : agent performing the action;
action : realisation of aordance;
mood: mood;
mood: mood;
mood : proposal | inducement | forecast | wish | palinode |contrition |
assertion | valuation;
time: time;
time: time.
The ST denition is recursive, which means that an ST can be related to an
agent within an ST structure. A minimum ST structure has three parts: an
agent, an action and a time, to be able to portray an agent performing an
action at a certain time. A minimum, rst-order ST normally refers to a
substantive action. It does not contain a modality but normally only a
proposition, and it sets up a direct relation between a sign and a substantive
action. A higher-order ST may have a reference to a lower-order ST; it
establishes a reference between signs. The modalities of a higher-order ST
are associated with the start and the nish at that level. This shows an
analogy with an ordinary use of language, that a speech or quotation can be
embedded into another speech.
The values of each particular mood are the illocutions (e.g. request,
command, permit, assert, assume, etc.). The value-types of time can be a
point-of-time, a period-of-time, and a relative time referring to other action.
Further denitions for both the modality and time will be given in
Appendix A, supplemented by examples and explanations. More discus-
sions about use of time can be found later in this chapter where the seman-
tic temporal database language, LEGOL, is presented.
10.2.2 ST for database design
The design of the structure of a database is set to satisfy requirements
derived from philosophical considerations and theoretical discussions.
Semantic properties, subjectivity, intentionalities and temporalities are
some of the most important aspects which should be entertained in the
semantic temporal database. In designing the structure of a database, the
ST has been taken as a basis for construction. Although, during the stages
of representing an informal system and implementing it in a technical
system, a great many properties of the informal systems may be lost, atten-
tion in devising the database design method has been paid to minimising
such expenditure.
A database record is called a surrogate. At the specication level, a surro-
gate is described as a tuple:
surrogate : surrogate-identication, content, antecedents,
authority, authority, mood, mood, action-time, action-time,
record-time, record-time
where
surrogate-identication provides a unique key for each realisation of an
aordance, by which one can equate the morning star and the
evening star;
contentcorresponds to the semantic part of an ST which represents a
semiological action (e.g. a speech act) or a substantive action (e.g.
being absent from school);
antecedents establishes links between the surrogate and its ontologi-
cal antecedent surrogates, which enables the association of the surro-
gate with the semantic context;
authority and authority relate the action to the responsible agents
who certify or recognise the start and nish of the action;
mood and mood indicate the intentions of the action represented in
the surrogate (the mood for start may dier from the mood for nish,
e.g., a surrogate may record a course of action started with a mood of
assertion and nished with a mood of retraction, in a case where the
agent realised what was asserted is a mistake);
action-time and action-time denote the start time and nish time of
the action;
record-time and record-time signify the start time and nish time
when the action is reported or observed.
Example Box 10.1: A speech act in a database
To illustrate the construction of a surrogate in a database, it may be con-
venient to continue with the example used in illustration of ST. The sur-
rogate for that example is specied as follows:
srg-id123, [absence], John, course123, auth123,
a b c d e
, assert, , 1-6-90, 30-6-90, [yesterday], .
f g h i j k l
The explanation of the surrogate is given below (the letters below the
surrogate label the items in order to relate to the explanations).
The speech-act agent, Susan, is not recorded in the surrogate; but if
it is necessary to have information about the agent, Susan can be
regarded as an authority responsible for recognising the action of
Johns absence. Normally, the responsible agents to certify or recognise
a type of action are generally dened in a problem domain (e.g. in an
organisation).
(a) Firstly, srg-id123 is the surrogate identication which is uniquely
generated by the database system.
(b) Next, [absence] is a dened action type (an aordance dened in
the ontology model) in the database.
(c),(d) Antecedents John and course 123 relate the action, being
absent, with the semantic context by adding the action agent and
action object.
(e) Next, auth123 is the authority who recognises that John is
absent. The authority, in this case, may be the teacher of Johns
class.
(f) The unlled authorityindicates that this record is valid, i.e., no
one has considered that this recording is a mistake needing to be
corrected. A lled authority, however, would indicate a respon-
sible agent for identifying the recording as an incorrect observa-
tion if someone as the agent does so.
(g) Next, assertion, as a mood for start, signies that the surrogate is
supposed to contain a reported fact or observation (not a
hypothesis or a prediction, for example).
(h) The unlled mood(i.e., mood for nish) designates that no one
disagrees with the surrogate; the observation is considered
correct.
(i),(j) Two times, 1-6-90 and 30-6-90, indicate the period of the action
of being absent.
(k) The time, [yesterday], tells when the observation of the action is
made. The database system may transform [yesterday] into a
proper time representation.
(l) The unlled record-nish-time indicates that the surrogate is still
valid. This eld of time must be lled in accordance with the
authority(the item (f)); that means the item (f) will indicate the
responsible agent who decides the record as being incorrect and
the item (l) records the time of his making of the decision.
10.3 Systems construction
After systems analysis and design, the next step is system implementation
in which the database records are dened.
A surrogate is a tuple in a database; it is a semiotic representation of an
aordance. A surrogatebase is a semantic temporal database containing
and managing a set of surrogates. A surrogate has a basic structure with a
xed number of intrinsic properties. Aordances in an ontology chart rep-
resent universal types whereas the instances of the aordances are values
of the universal types. Behind the ontology chart, all the universal
aordances have the same number of properties, such as surrogate-
identication, label, antecedents, etc., which correspond to the compo-
nents indicated in the surrogate denition. For a system design, what is
required is the specication of the property values of each aordance
identied in the ontology chart. Data on the particulars have the same
structure as the universals. With regard to the example of project manage-
ment (see Figure 6.5), data containing the universals and the particular
instances can be as shown in Table 10.2 which illustrates the major proper-
ties of the surrogates.
The implication of this canonical structure is that once a semantic model
is produced, the design of the database is already virtually known because
the components of the surrogate are predened. If there is a computer
supported environment available for capturing the semantic model and
Table 10.2. Database excerption of the example of university
administration.
sr9-id label ant1 ant2 type action action record record
xx1 organsn xx0 xx0 u
xx2 Sta Uni. xx0 xx0 xx1 19600901 19980901
xx3 Twente U. xx0 xx0 xx1 19650901 19980901
xx4 department xx1 u 19980901
xx5 Computing xx2 xx4 19900901 19980901
xx6 person xx1 u 19980901
xx7 Piet xx2 xx6 19700203 19980901
xx8 Peter xx3 xx6 19230205 19921231 19980901
Note:
action+, action, record+ and record are the start/nish times for
action/record, equivalent to the notions of valid time and transaction time in temporal
databases.
The type having a value u indicates the surrogate to be a universal, otherwise the
surrogate belongs to the type of the surrogate referred to in the eld type.
translating the model into the database design, the implementation of the
database can be immediately realised. This would allow one to concentrate
on systems analysis, and would make it possible to obtain a prototype
system for testing and verifying the result of systems analysis. When the
verication is satisfactory, the system can be used as an information system
for real business applications.
10.4 LEGOL
The project LEGOL was originated in the early 1970s (see Stamper (1980)),
one of the objectives being to develop a language, with the same name as
the project title, for specifying legal rules and also for operating on rela-
tional databases (Jones et al. 1979). Several versions had been developed
earlier. LEGOL-3.0, based on the early versions, is devised for two pur-
poses: as a specication language for norms and as an operation language
for the semantic temporal databases. This section will give a brief introduc-
tion to LEGOL.
10.4.1 Basic syntactic structure
The statements in LEGOL can be seen as norms specied in a formal
syntax. The norms can dene database constraints, specify applications in
the databases, and trigger human actions. The basic structure of a LEGOL
statement is the same as that of a norm. There are two parts composing a
statement:
condition consequence.
The condition part species the circumstances in which the actions will be
taken, while the consequence part designates the actions. It is possible to
have just the condition part, then the statement will be seen as a query in the
database.
There are two types of actions dened in LEGOL for the consequences:
one type is called substantive and the other semiotic. Two specic actions
are dened as the substantive kind, start and nish; and two as the semi-
otic kind, print and report. They can be schematically grouped as in
Figure 10.1. Each tuple stored in the database is a representation of a par-
ticular of an aordance which has a start time and a nish time associated
with it. The operations of the conventional (relational) databases such as
insertion, update and deletion can be suciently realised by these two
operators. The start and nish deal respectively with the start time and
10.4 LEGOL 143
the nish time of the existence of an aordance: When a tuple is entered as
a database record, a start time and a nish time will be lled to indicate the
existence of the particular aordance (as the event times, see Figure 10.1).
The start of the transaction time is lled with the current time of the opera-
tion on the database. When a tuple is deleted, the tuple will not be physically
erased from the database, but a nish for the transaction time will be lled
with the current time of operation on the database. The two operations, to
start and to nish, will actually change the states of aordances; that is why
they are called substantive actions. The semiotic operations are done by the
operators print and report. The former will direct the database output to
be displayed onto the screen, and the latter will send information into a disk
le to generate a report.
To demonstrate the basic shape of the LEGOL statements and the conse-
quence operations, the following norms for the CRIS case (see Appendix B
for the ontology model) are formulated below and the corresponding
expressions in LEGOL are supplied in Table 10.3. More examples and
query results can be found in Appendix B.
Norm 1: Any contributor of an abstract or member of TC or WG asso-
ciated with the conference is eligible for the conference.
Norm 2: Any member of the OC, PC or any author of a selected paper
has priority 1.
Norm 3: An invitation should be sent to the author within two weeks
after his paper is selected.
As can be seen from the examples, the consequence part of the statements
contains action operators, e.g. start, nish, and report; the condition part of
the statements may involve many operators such as while, while-not, or-
while, and so on. Some operators used in the condition part of the LEGOL
statements will be discussed in the next sub-section.
action
substantive
start
finish
semiotic
print
report
Figure 10.1. Types of the consequent actions.
10.4.2 Some important operations
Most of the operators appear in the condition part of LEGOL statements.
If the consequence part is not explicitly mentioned, then the statement is a
query and the result will be displayed on the default output terminal (i.e. the
screen). The LEGOL operators involve two kinds of operations: a selection
from the database contents and a calculation on the basis of time.
Examples of the operators having explicitly strong relations to time are
while, while-not, after, within . . . after; the ones having implicit relations to
time are whenever and whichever. Another way to classify the operators is to
put them into two groups according to the syntactic features, the unary
operators and the binary operators, which will be a line for the following
discussion on some LEGOL operators.
2
Binary operators
The syntax of a binary operator is operand binary-operator
operand. Only three binary operators will be discussed below; many
10.4 LEGOL 145
Table 10.3. Examples of LEGOL statements.
1. (contribution(person, abstract) while on(subject(WC#CRIS-3), abstract))
or-while (membership(person, TC) while sponsors(WG(TC), WC#CRIS-3))
or-while (membership(person, WG) while sponsors(WG, WC#CRIS-3))
start eligible(person, WC#CRIS-3).
or using role names:
(contributor(abstract) while on(subject(WC#CRIS-3), abstract))
or-while (member(TC) while sponsors(TC(WG), WC#CRIS-3))
or-while (member(WG) while sponsor(WC#CRIS))
start eligible(person, WC#CRIS-3).
2. membership(person, OC(WC#CRIS-3))
or-while membership(person, PC(WC#CRIS-3))
or-while (contributes(person, paper) while selected(paper))
while on(paper, subject(WC#CRIS-3))
start priority(eligible(person, WC#CRIS-3)) = 1.
3. (contributor(paper) while on(paper, subject(WC#CRIS-3)))
within 2 week after start-of selected(paper)
report invitation person, WC#CRIS-3.
and the control norm:
(contributor(paper) while on(paper, subject(WC#CRIS-3)))
while start-of selected(paper) while-not invitation(person, WC#CRIS-3)
report invitation-delay person, WC#CRIS-3.
2
In fact there are more LEGOL operators in addition to these two categories, e.g. between
. . . and . . ..
others, e.g. when, whenever, whichever, after, before, may directly appear in
the text because they may even be understood without detailed explana-
tions.
operator: while
syntax: x while y
semantics: The x and y are aordances. The result is a complex type
expressed in a tuple (x, y, event-start, event-nish). The
interval indicated by the event-start and the event-nish
results from the time intersection between x and y.
example:
3
person while membership(person, TC)
The question in natural language is who, during what
period, is a member of a TC.
This query will show all the persons who are members of a
TC. The event-time for each person will be the period while
he is a member of a TC.
operator: or-while
syntax: x or-while y
expressed in a tuple (x, y, event-start, and event-nish). The
results from the time union between x and y.
example: membership(person, TC#2) or-while membership(person,
TC#8)
The question in natural language is who, during what
period, is a member of either TC2 or TC8.
This query will give a result of the persons who are
members of either TC2 or TC8. The result contains the
following items: person, membership, TC, event-start,
and event-nish.
operator: while-not
syntax: x while-not y
expressed in a tuple (x, y, event-start, and event-nish). The
results from the time exclusion of the existence of x by that
of y.
example: national-org while-not membership(national-org, IFIP)
The question in natural language is which national
3
Examples in this sub-section are based on the CRIS case; see Appendix B for more examples.
(information processing) organisation after its creation,
during what period, is not a member of IFIP.
This query will produce a list containing the national
information processing organisations that were not
member organisations during certain periods. The listed
organisations must now be IFIP member organisations
which did not become IFIP members immediately after
their creation.
Unary operators
A unary operator has an operand on the right. The syntactic usage of a
unary operator is: unary-operatoroperand.
The three operators, current, past and future, select the particulars of the
aordances in dierent states. The event start and the event nish times of a
particular instance indicate one of the three states that the particular
belongs to (as shown in Figure 10.2). The operator current will select the
tuples in the database which have a start time in the past and a nish time in
the future (with known or unknown exact times). The operator past will
select the tuples with both a start time and a nish time in the past; and the
operator future with a start time and a nish time in the future. These oper-
ators can be put in dierent places, hence have dierent eects. See the fol-
lowing LEGOL statements for examples.
10.4 LEGOL 147
Temporal location of affordance Operator
present
(the moment of the operation)
past
past
current
current
current
future
future
Figure 10.2. Temporal locations of aordances and operators.
Example Box 10.2: LEGOL statements
membership(national-org, IFIP) (1)
current membership(national-org, IFIP) (2)
The operators of start-of and nish-of will produce the event time of the
start and the nish of the aordances. For example, start-of
membership(national-org#NGI, IFIP) will return a value of time 1962-01-
04 supposing NGI was admitted by IFIP as a member on January 4, 1962.
The operators before and after are dened as unary operators (notice they
can be used as binary operators, e.g. WG(IFIP) after WG#WG8.1). If an
expression in LEGOL is
before start-of membership(national-org#NGI, IFIP)
the result of this statement can be a time interval up to the time when the
organisation NGI became a member of IFIP, i.e. , 1962-01-03.
Functions
There are functions dened in LEGOL such as arithmetical calculations
and analytical functions. The analytical functions are grouped into two cat-
egories: diachronic and synchronic. The diachronic functions operate
membership(current national-org, IFIP) (3)
membership(past national-org, IFIP) (4)
future membership(national-org, IFIP) (5)
membership(future national-org, IFIP) (6)
The result of the execution of statement (1) will be all the national
organisations which were, are and as far as we know will be IFIP
members.
Statement (2) will give all the present membership of the IFIP
members.
Statement (3) will show all the periods of membership in past, present
and future of the current members (if a national organisation had
been expelled from IFIP and then some time later re-admitted, this
would appear as two separate instances of membership).
Statement (4) means the memberships of the national organisations
which no longer exist.
Statement (5) shows which organisations IFIP expects to admit as
members.
The last statement would show, if possible, that IFIP will (or may)
admit a national organisation which will (or may) be founded in the
future.
across the time horizon while the synchronic ones calculate the result with
respect to the distribution over time. For example, the use of a diachronic
function as in count member (IFIP) will summarise all the IFIP members
over the complete IFIP history. The counted value will not decrease but
only increase whenever there is a new member joining IFIP. The counter-
part of count is the synchronic operator number. The use of number
member (IFIP) will produce a list of present IFIP members distributed
over the history.
The number will increase whenever there is a new member joining IFIP,
and decrease when a member has left IFIP. The time periods in which the
two types of functions calculate can be specied by operators associated in
the same expressions. Table 10.4 summarises the analytical functions
dened in LEGOL.
10.4 LEGOL 149
Table 10.4. LEGOL functions.
Diachronic functions Synchronic functions
count number
accum sum
highest/greatest max
lowest/least min
11
Normbase: a new approach to information
management
Normbase is not just a software system. It represents a new way of manag-
ing information. It also represents a new approach to the development of
information systems using semiotic methods. The chapter will introduce
the concept of Normbase, and the Normbase system as a software environ-
ment for managing information. It will also discuss how the Normbase
system can be used for supporting business management and decision-
making.
11.1 The Normbase concept
Norms are parts of business knowledge that determine the meanings of the
data in databases. The knowledge serves as a constraint upon the organisa-
tion and utilisation of the data. In order to allow only meaningful interpre-
tations of and operations on the data, the knowledge must be consolidated
into the technical information systems in a proper way.
The database discipline has established an approach to incorporating
business knowledge into the technical information systems following the
principle of data independence. Data independence, as one of the objects of
database systems, provides an important separation between data and
applications. It furnishes an immunity of application programs from
changes of the data storage structure and access strategy (Date 1995). Data
independence is achieved by isolating the domain-specic (knowledge)
parts of the database system from the general (supporting) part. This
process of knowledge abstraction can be applied not only to databases, but
also to programming techniques. In object-oriented programming lan-
guages, the technique of abstract data types allows one to specify certain
operations on a data type without having to aect the internal represen-
tation of the instances. In databases, the domain-specic knowledge is
150
represented in a conceptual model, which is typically stored in a data dictio-
nary. Practice of database development normally follows the principle that
the more knowledge is put in the conceptual model, the less remains for
the application programs (Weigand 1990).
Stamper et al. (1991) advocate a higher-level data and knowledge inde-
pendence in information systems: a liberation of business knowledge from
the application programs. After the evolution of techniques of information
systems development being reviewed, a concept of the Normbase has been
introduced in their paper. The Normbase structure will be created, as it
states in the paper,
. . . that will complete the separation of data management from applications pro-
gramming, extend the distributed systems concepts, and also divide applications
programming into parts that focus exclusively on knowledge of the organisation
(expressed by the norms of teams, rms, social groups or nation states) or upon
knowledge of how to exploit the technology.
The emergence of the Normbase system suggests a strategic advance in
information systems development, which can be schematically summarised
as in Figure 11.1. With the present database approach, business knowledge
is interpreted as program semantics and reected in application programs.
The underlying part, DBMS, is only concerned with syntactic constraints,
typically data normalisation. The database approach eliminates the dedica-
11.1 The Normbase concept 151
Application
4GL (e.g. SQL forms)
RDBMS
Database paradigm
Normbase paradigm
Migration
Application
Normbase engine
STDB Norm store
Constraint
implemented
in procedures
Limited operational
constraints and
simple triggers
Mainly technical
constraints, e.g. data
normalisation
Application: only operational commands
(e.g. report formation); business
knowledge organised in the Normbase
Normbase engine: maintaining and
executing norms
STDB: storing semantic temporal data,
constrained with ontological dependencies
Normbase: managing norms, constraints
imposed on the data and applications
Figure 11.1. From database to Normbase.
tion of the data to specic application domains; however, the application
programs remain problem-specic. This means that any newly emerging
application requirement implies a new application to be programmed; any
modication of the requirement embedded in an existing application
module may require an alteration in the application software.
This Normbase approach requires, rst of all, an information model
which reects the most basic knowledge of the social world. The ontology
model describes the most fundamental patterns of behaviour. The scope of
the model is determined according to the requirements of the users of the
system. In relation to the information model, the norms are elicited by
studying the agents in action. The norms will be written formally and
managed in a centralised knowledge base, the norm store. The data are
organised according to the ontology model and stored in the surrogatebase.
The standard database operations are available to the surrogates. The
norms centrally managed in the normbase are treated in the same way as
data but in a more complex structure. Standard operations, such as inser-
tion, update, etc., are possible on the norms as well. The links between the
surrogates and the norms are established and maintained by the Normbase
engine. When any operation in the system takes place, the relevant norms
will be invoked to check the consistency and validity.
The Normbase approach is not just a modied technical solution. It
takes a revolutionary step toward a completely new paradigm of informa-
tion management. It requires one to adopt the viewpoint that an organisa-
tion is an information system where people act within the constraints of
norms. The objects managed in the technical information system are not
just objective and timeless descriptions of the world, but are representa-
tions of agents in action based on subjective viewpoints. The Normbase
system is an extension of the capacity of human agents which should
always incorporate their changeable personal values and judgements.
11.2 The Normbase system
The Normbase system is a software environment for management of infor-
mation and norms. These are the major software components of the
Normbase system: the Normbase engine, the Semantic Temporal
DataBase (STDB) and the norm store.
11.2.1 The Normbase engine
The Normbase engine is composed of a semantic analyser, a Norm
Analyser, a LEGOL interpreter and other data management facilities.
152 Normbase
Semantic Analyser
Semantic Analysis is the rst step in setting up a Normbase for a particular
application. The Semantic Analyser is a facility to assist a user to produce
an information model and to design the database. The Semantic Analyser
provides a knowledge-based semantic charter, which remembers and
applies the knowledge of Semantic Analysis. Together with the description
of the semantic units generated by a concordance generator, the Semantic
Analyser can perform semantic checking on whether a user builds an
ontology model correctly. The concordance generator is a specially
designed text-processor for Semantic Analysis that takes a piece of text (a
problem description) as input to produce a list of semantic units. One can
then give descriptions to each semantic unit, for example whether it is an
agent or aordance, and whether it is a role and in what relation this role is
involved.
The Semantic Analyser opens an ontology chart editor window for a
user. With the Semantic Analyser, the user can draw graphic entities on
the ontology chart, link the lines between the entities, and modify and
move the entities around in the window with the lines connected. The
Semantic Analyser can optimise the connection lines if the user wishes. It
also allows the user to edit the graphic entities and lines manually. A user
can get the ontology chart created previously from the system storage
onto the window. The user can also let the Normbase system produce a
new ontology chart automatically with optimised positioning of the
graphic entities.
Norm Analyser
Norm Analysis is the second important step in setting up a Normbase
system for an application. The norms will be captured and represented
through the norm analyser. In most cases, the norms are linked to the uni-
versals (universal concepts), though sometimes norms will have to be linked
to particulars (particular instances). The norms to be captured in the norm-
base system are classied into three kinds.
Constraint: dening status or relations other than ontological depen-
dency. These norms specify the data consistency and legitimacy from
the business point of view.
Authority: dening authorities for the start and the nish of an
aordance. The concepts of responsibility and authority are closely
related. Unless the responsibility and authority are clear, a business
can hardly be managed successfully.
Trigger: dening conditions and actions to be taken by responsible
agents. Actions will be taken by the Normbase system as soon as the
conditions are met, or messages will be produced to call for human
actions.
LEGOL interpreter
The interpreter accepts LEGOL statements and translates into a series of
appropriate actions. A LEGOL statement usually contains two parts. The
rst part supplies the conditions and the second part species a consequent
which often results in an action. The actions can be as follows.
Retrieval of information according to requirements stated.
Evaluation of the state of aairs The interpreter takes into account the
facts supplied in the statement and records in the system. The result of
evaluation can be advised to the user immediately or recorded in the
system.
Generation of an event The event can be producing an email message and
sending it over the network, or sounding an alarm.
Notication, i.e. a message displayed on the users screen.
11.2.2 The semantic temporal database
The Semantic Temporal DataBase (STDB) manages the data for the
Normbase. It performs standard database operations, such as insertion,
retrieval and update, and other services to the Normbase.
Generation of a database structure from a semantic model The semantic
model generated by the semantic analyser is usually in the form of an
ontology chart, though sometimes it can be a schema (a textual
description). Once the semantic analyser has produced an ontology
chart, it will be used as the conceptual model from which a database
structure is generated automatically. The STDB has a uniform record
structure, as discussed in Chapter 10. An illustration is given in
Example Box 11.1.
Data management. STDB manages data according to the semantic
models. The ontology chart denes ontological relationships between
antecedents and dependants. This ontological consistency will be all
the time maintained by STDB, and will be used to check every data-
base operation, e.g. insertions and updates. All updates are non-
destructive, which is achieved by using the concepts of action times
and record times.
154 Normbase
11.2.3 The norm store
The norms are kept in the norm store. The same as the data in the STDB,
each norm has a unique identity number. Each also has a start time and a
nish to identify the validity period. Norms can be inserted and updated, in
a non-destructive way, as data. Norms function as conditions and con-
straints for the system operations. All the norms are coupled with appropri-
ate aordances in the STDB. The functions of the norm store are to
Example Box 11.1: Generation of an STDB from an ontology chart.
This example box shows a simple ontology chart (Figure 11.2). The
ontology chart is then used as the basis to generate a STDB structure.
See Table 11.1.
Only certain elds are shown in the database for illustration purposes.
The surrogate identities (sur-id) are generated by the STDB as unique
keys to each record. The labels contain the three concepts in the ontol-
ogy chart. All three records are universals, indicated by u in the eld
type, one agent person and two other aordances, as indicated in the
eld sort by a and af. The antecedent elds (ant1 and ant2) tell that the
concept holds ontologically depends on person and book. All start times
of action (act) are zero, which means these concepts came into exis-
tence a long time ago. The unmarked nish times of action (act) indi-
cate these concepts are still valid (as opposed to a known nish time
which means an invalidity).
person
holds
book
Figure 11.2. From database to Normbase.
Table 11.1.
srg-id label ant1 ant2 type sort act act
1 person 0 0 u a 0
2 book 0 0 u af 0
3 holds 1 2 u af 0
maintain the consistency of the norms and invoke appropriate norms
whenever the Normbase performs an action.
11.3 Information management with the Normbase system
The Normbase system is a software environment to support the semiotic
approach to information systems development. It enables one to conduct
activities at all stages of the development of an information system.
Semantic modelling and Norm Analysis The semantic analyser assists the
user to conduct a Semantic Analysis, starting from a problem descrip-
tion, to an ontology chart. The semantic analyser, following the princi-
ples and steps of the method of Semantic Analysis, prompts
appropriate templates and questions to guide the user through an
analysis. It performs all kinds of checking during the analysis, such as
ontological consistency, genericspecic, role, and so on.
The norm analyser provides an environment for specifying norms
and entering norms into the norm store. It also enables one to specify
156 Normbase
Example Box 11.2: Norms in the norm store.
Suppose we are in the context of a person borrowing books from a
library, this is a norm (a membership regulation).
If a person already holds ten books, then that person is not allowed to
borrow more.
This can be translated into an expression close to LEGOL:
Whenever count(holds(person, book)) 10,
The person is not allowed to borrow.
In LEGOL, it reads like this:
Whenever count(holds(person, book)) 10
Forbidden to start holds(person, book)
The boldface words are LEGOL operators or system reserved words.
The LEGOL operator count is a synchronous operator which will check
at this point of time how many instances of holds(person, book) a par-
ticular person has. If it reaches 10, then the system will forbid us to
insert (i.e. to start) a new record of holds(person, book).
This norm is coupled with the concept holds in STDB. This entails
that every time there is an operation to start holds(person, book) in the
STDB, this norm will be invoked to check the legality.
the coupling between the norms and relevant agents and aordances
in the STDB.
Populating STDB Once a Semantic Analysis is completed, a STDB will
be generated from the ontology chart. To populate the STDB, the
Normbase system provides a multifunctional browser by which one
can enter instances of aordances into the STDB. As determined by
the nature of the STDB, to insert a record of a particular aordance is
to start an instance of that aordance, while to delete a record is to
nish an instance of an aordance.
Information retrieval One can retrieve information from the Normbase
system with the multifunctional browser. The browser can be used for
data entry as well as data retrieval. Another way to get information is
to write commands in LEGOL. In this case, one can write very sophis-
ticated expressions using time-bounding operators such as while, or-
while, while-not, and other functions. However, for users who do not
wish to write queries in LEGOL, there is a query-by-table facility
where the users can select items and then specify criteria in a user-
friendly manner.
Management and decision support The norms in the system can function
as triggers; therefore the system can proactively support management
and decision-making activities. The Normbase checks every opera-
tion, internal and external event. The events include input from a user
or from another connected system, advance of a time clock, and
change of the value of a record. The system evaluates the norms
against the current state of aairs in an organisation, and can take
actions or prompt relevant agents to act if necessary. In this way, the
Normbase can be used as an assistant to a manager, or can be dele-
gated to control some business processes.
11.4 Using semiotic methods with other approaches
In a project of information systems development, one can use the semiotic
methods presented in this book in conjunction with other methods and
approaches. As one may have experienced with object-oriented methods, a
project can use methods of OO analysis and design, but use non-OO pro-
gramming languages to implement the system (for example, C). The
reasons for doing this may be that these OO methods bring some benet in
analysis and design, but a non-OO programming language may result in
operational eciency. Nevertheless, most of the OO methods are language
independent, which lends the possibility of choosing any conventional lan-
guage for implementation.
The semiotic methods can be used for systems analysis and design, and
then followed by other approaches. Or, even, one adopts the semiotic
methods for analysis only, then applies OO or other approaches. If a
process of IS development is identied as activities in four stages, these are
viable permutations (Table 11.2).
Option 1 adopts semiotic methods to cover all the stages of IS develop-
ment, which involves the Semantic Analysis Method (SAM) and Norm
Analysis Method (NAM) in the analysis and design stages and the
Normbase system in the implementation. Other options involve semiotic
methods in various early states and other methods later. It is worth men-
tioning that the semiotic methods such as Semantic Analysis and Norm
Analysis will give a great added value to an information system project even
if they are only used in the analysis stages. This is because these semiotic
methods can help one to understand and articulate the requirements better,
and produce systems analysis in a more rigorous way. The requirements
and systems analysis are essential to the success of an information system
project, as discussed early on in this book.
11.4.1 Relational database for implementation
The function of a semantic model is not only for constructing an informa-
tion system. The main strength of the semantic modelling is to represent
158 Normbase
Table 11.2. Combining semiotic and other methods.
IS
development
activities Option 1 Option 2 Option 3 Option 4
Requirement SAM, NAM SAM, NAM SAM, NAM SAM, NAM
analysis
Systems SAM, NAM SAM, NAM SAM, NAM other methods,
analysis e.g. OO or
structured
analysis
Systems design SAM, NAM SAM, NAM other methods, other methods
e.g. OO or (see the entry
structured on the left)
design, ER
Systems NB other methods, other methods other methods
implementation e.g. OO or (see the entry (see the entry
other languages, on the left) on the left)
CASE tools
the users requirements in a precise and formal way so that the undesired
omission and misunderstanding of the requirements can be detected and
avoided. After an information model is obtained with the help of Semantic
Analysis, one of the most critical parts of a system development project is
completed. Therefore, a signicant benet is still gained even if one contin-
ues the design and implementation with other methods.
For demonstration purposes, a computerised prototype information
system has been built using ORACLE forms for the CRIS studies.
Surprisingly, the semantic model could be almost directly implemented
with the relational database tools though some slight modications were
needed. The database of the prototype was composed of ve relational
tables (Thnissen 1990) so that the features of the canonical surrogatebase
are accommodated. Four tables out of the ve are worth mentioning below:
the surrogates containing the intrinsic properties;
the relations retaining the ontological dependencies between the
aordances;
the norms storing the norms specied in English and LEGOL;
the agenda holding the triggering norms for actions to be taken in the
future.
Another example of using relational database technology in combination
with the Semantic Analysis is the administration system in the University of
Qatar (Ades 1989). The systems analysis work on the computerised admin-
istration system was done with Semantic Analysis and the implementation
was fully on a relational database language of WANG. One of the advan-
tages of the approach adopted in the project is reportedly the complete cov-
erage of the requirement and the clear specication.
11.4.2 Object-oriented methods for design and implementation
A natural alternative way of systems construction is the object-oriented
(OO) approach because of the similarities in viewing objects in the OO
approach and in handling aordances in the other.
Object-oriented design
Principles of object orientation were rst conceived in programming lan-
guages for systems implementation, and they have been further adopted
into the areas of analysis and design. The advantages of object orientation
in systems design and implementation have been claried in many works
(e.g., Booch (1994), Coad & Yourdon (1990), Micallef 1988)). But as
pointed out by van de Weg and Engmann (1992), the techniques and
underlying principles of object-oriented analysis and design are not well
understood and are still at an immature stage. To an extent, this observa-
tion still remains valid, although there have been some new developments
in analysis and design methods recently (Booch 1994, Coad & Yourdon
1990, Embley et al. 1992, Shlaer &Mellor 1988, 1992). Arecent most excit-
ing OOmethod is UML (Booch et al. 1998). As a result of combined work
froma number of authorities in the OOeld, UML oers a set of compre-
hensive techniques and supporting tools for object-oriented systems
design. But UML never claims to be particularly strong in user require-
ments analysis.
Some insuciencies in analysis methods have been observed in the OO
methods. For example, Coad and Yourdon (1990) oer a clear method of
analysis which consists of ve major steps: identifying objects, identifying
structures, dening subjects, dening attributes and instance connections,
and the last step, dening services and message connections. But the
method does not give a sucient account of how to elicit and represent user
requirements in a concise and understandable form for the user. Most
analysis methods are based on an assumption that there is a given list of
user requirements, so the task of analysts is to represent the requirement in
data models. The most often used modelling methods for object-oriented
systems development are the extended ER method and structured
method.
1
These approaches to requirement analysis and modelling put a lot
of emphasis on logical data organisation. But they do not question at all
these important issues: Are the requirements given by the user correct? Is
what is said what is really wanted? Is the understanding of what is said
correct? Is there is a twist or distortion that may be caused by the analysts
misinterpretation?
The Semantic Analysis method takes especially into account the problem
of language semantics that may arise during the requirement elicitation and
analysis. The method provides a formal way of representing user require-
ments in a succinct form which can be easily understood by the user and
subjected to the users critical examination. Therefore, verication at the
analysis stage can be realised and this will, to a large extent, guarantee a
fundamentally sound basis for the rest of the work in the system develop-
ment lifecycle. The object-oriented design can be made on the basis of the
analysis result. For this purpose a set of transformation rules is provided so
that an object-oriented design can be smoothly derived from the result of
analysis, and further constructed using object-oriented programming tools.
160 Normbase
1
The ER method can be best represented by Chen (1976), the structured method by
Yourdon & Constantine (1979) and Yourdon (1989).
From Semantic Analysis to OO design
To conduct a Semantic Analysis is to capture and articulate user require-
ments. The semantic model produced from the analysis tells what objects
are needed and the semantics of the objects. However, the model does not
readily lend itself to implementation. It needs a transformation in order to
derive an object-oriented design.
The three principles of transformation from a semantic model to an OO
design are as follows:
Principle 1: All information in a semantic model must be used in derivation
of a design. Any alteration of the semantic model must be approved by
the user.
The semantic model reects a conceptualisation of the problem. It
results from the consensus between the analysts and users. Sometimes
there may be more than one user group involved in the project coming
from dierent professional backgrounds. Therefore, the semantic
model serves as a kind of document with authority for the following
system development activities.
Principle 2: There are mapping rules between terms in the semantic model
and the object-oriented design:
agents, entity-like aordances objects;
determiners attributes;
actionlike aordances communications between objects;
roles attributes and static subset
constraints;
whole-part relation nested or separate objects
(depending on the programming
language);
genericspecic object inheritance.
The semantic denition of each aordance is determined in the
semantic model. The denition includes attributes and actions of an
object, and interactions between objects. All information contained in
the semantic model can be useful in producing the design model. For
example, a role name in the semantic model would suggest a synonym;
in an OO design, it can be treated as an attribute if the synonym is
required. It also suggests a subset of the role-carrier, because, e.g.,
teachers consist of a subset of persons. Roles are dened by certain
rules and are represented in the OO design as constraints.
Principle 3: Norms associated with the semantic model must be satised as
conditions and dynamic constraints for actions between objects.
Some actions should be taken if the conditions are met. These con-
ditional norms are called triggers. The norms are prescriptive in the
sense that they instruct agents when and what actions must, may,
and may not be performed. These norms will be used as constraints
incorporated into objects for controlling their behaviour.
The design of the system is basically meant here as the design of a single
program which is similar to an Ada or VMS task or a Unix process, as the
same denition is used by Shlaer and Mellor (1992). They suggest that there
are four signicant aspects of design which are class diagram, class struc-
ture, class dependency, and class inheritance. This general direction has
been followed in developing this design method. However, much
simplication has been made for practical purposes. In the proposed design
method, three major aspects are dened, which are class description, inter-
class communication, and inheritance. An introduction to these three
aspects will be given in this section (illustrations of the application of the
method will be found in Chapter 13 on the CONTEST system develop-
ment).
A class description consists basically of logical components, functions,
and state transitions. The logical components of a class are dened using the
class template which is composed of a class-name, brief explanation of the
functions, visibility (exported/private/imported), cardinality (0/1/n), hier-
archy (super-classes, meta-class), interface, implementation, lifecycle, etc.
The specication of the interface must indicate authority over the attrib-
utes and member functions by specifying whether they are public, protected
or private, and stating the clients and servers of the classes. The functions
162 Normbase
Icon for a class Icons for class relationships Cardinality
A B
A uses B (for interface)
A B
A uses B (for implementation)
A B
A is inherited from B
name
0 zero
1 one
* zero or more
+ one or more
? zero or one
n n
Figure 11.3. Design notations (based on Booch (1994)).
are specied with the following elements: name, nature of the function
(public, protected or private), input and its data-type, output and its data-
type, the states of the class before and after the invocation of the function,
other classes involved in performing the function (client and server classes),
etc. The states of a class are specied by the state transition diagrams on
which a class can be led from one state to another by invoking associated
actions. The actions are normally reected in functions, either the member
functions of the class or friend functions of other classes.
There is a variety of relationships that are meaningful among classes,
including inheritance, using, instantiation, and meta-class relationships
(Booch 1994). The most important relationship is communication which
can be modelled with inter-class communication diagrams. Figure 11.3
shows some example notations for modelling relationships. In an inter-class
communication diagram, the two A uses B notations are employed to indi-
cate the using relationship in terms of sending messages by invoking certain
functions. The function names will be placed along the linkage lines. For
example, the diagram in Figure 11.4 means class A uses class B by invoking
function x, and B is used in As implementation. The ways of invoking
functions should be specied in the class description.
The class inheritance is illustrated by diagrams to show inheritance rela-
tionships between the parents and children classes, for which the arrow is
used, as in the notation shown in Figure 11.3. Logical components of both
the parents and children classes designed with the class template should be
used in association with the class inheritance diagram in order to know
exactly the inherited part.
Once an object-oriented design is completed, the implementation can be
done in any object oriented, or even any conventional, programming lan-
guage.
A B
function_x
Figure 11.4. Class A uses Class B.
12
Case study: development of a land resources
information system
In this chapter, the application of Semantic Analysis, Norm Analysis and
the Normbase system in the development of a land resource information
system oers an example of adopting the semiotic approach to cover activ-
ities of all stages of information systems development.
1
12.1 Background
The assignment was originally for students in a postgraduate GIS (geo-
graphic information systems) course at the International Institute for
Aerospace Survey and Earth Science, the Netherlands. The course partici-
pants, after having worked in dierent disciplinary elds for many years,
came from all over the world. They were supposed to use some structured
method to carry out the systems analysis and design. The purpose of apply-
ing the semiotic methods was originally for a simple comparison between
dierent approaches. However, for the purposes of this chapter, the analysis
and design with the other approach will not be presented.
The objective of the case is to develop an information system for man-
agement of land resources, for an imaginary country, Snake Island. This
project represents a typical type of information systems that are currently
in high demand in many countries, especially the developing countries.
Developing such a system requires tremendous eort, because very often
such a system involves a heavy investment on expressive GIS hardware and
software, and a costly multidisciplinary team of experts. On the other hand,
such a system normally has a signicant political, economic or other
impact on the region that the system is supposed to serve. Therefore it is
important to have an eective methodology for system development, cover-
164
1
Part of this chapter was presented by Liu and Stamper in a workshop in GIS in Wuhan,
1991.
ing requirement analysis and specication, system design and implementa-
tion.
Brief description of the problem
Snake Island is an imaginary beautiful country situated in the middle of the
South Pacic. It has bare mountains, green hills, agricultural plateaux,
ood hazard areas, and smooth beaches. The land resources are mainly
used for urban, agricultural, and tourism purposes.
Due to the fast rate of economic development, the Land Register oces
encounter great diculties in timely and correct registration of rights on
land and buildings for planning, management and taxation. Therefore the
central government of Snake Island formed an inter-departmental com-
mission to develop a Multi-purpose Cadastral Information System. The
proposed information system has the following objectives:
(1) to help the Land Registers to keep the information about the ownership
of land and buildings up to date;
(2) to automate the preparation of homogeneous zoning maps for the
Department of Valuation of Agricultural Land;
(3) to help the Land Tax Department in preparing the tax bills to be sent
to all owners of the land to be taxed;
(4) to assist in preparing information for the Warning and Evaluation
plans required by the Safety and Emergency Department.
12.2 Semantic Analysis for requirements modelling
The Semantic Analysis method is used to produce semantic models (or
ontology charts).
Starting point the problem denition
The analysis normally begins with a written problem denition. The
problem denition may sometimes just be written not precisely enough. It
will be partial, usually with a measure of vagueness, and not infrequently
containing contradictions. A set of techniques for problem articulation
can be employed to extend the problem denition (see Kolkman (1995)), if
it is necessary. It is rare to nd a satisfactory problem denition in a written
document.
Study problem denition
Semantic Analysis must begin with such a defective problem denition as
an input le. But it is recommended in this step to have sucient assistance
from users for understanding the problem. The method of analysis system-
atically leads to the clarication of the problem and it increases the mutual
understanding among the users. There are criteria for identifying progres-
sively larger swathes of the problem for analysis. For example, initially
attention is focused on the substantive aspects of the problem, that is,
excluding the issues of control and message passing. The whole procedure
will be iterated, gradually increasing the scope of the problem being exam-
ined.
In our study case, the case document was the problem denition, where
the nature of the problem, system users, and requirements are stated.
Identify semantic units
This step is something like lexical analysis on the problem denition. In this
step, the problem denition is taken as input; the output is a list of phrases
having specic meanings in their context. This is the list obtained after
having conducted the analysis at this stage:
province
district
village
person
owner
ownership
land (incomplete knowledge)
tax (incomplete knowledge)
Department of Valuation of Agricultural Land
Safety and Emergency Department
Cadastral Survey Department
Department of Land Tax
. . .
As listed, there is incompleteness of knowledge about some semantic units,
which suggests some more eort on eliciting knowledge in these aspects.
For example, land is such an item on the list. It seems obviously impor-
tant. But the precise meaning of land is not so clear; what information
about land should be modelled may not be fully understood. To know
these, more knowledge about land should be obtained either from study-
ing available documents or from interviewing the problem-owner.
After some more eort, more knowledge about land is gained.
Therefore, the following list of semantic units can replace land on the pre-
vious list:
166 Case study: land resources information system
land-unit
parcel
agricultural parcel
urban parcel
. . .
And the following list can replace tax:
tax-rate
tax-bill
Classify semantic units
The terms on the list can be analysed in several categories.
Agentaordance In a social system, all phenomena can be classied into
two categories: agents who can do actions themselves, and aordances
which are actions or results of actions of those agents. The agents con-
struct the whole social world; the agents can hold responsibilities for
their actions.
Universalparticular The concept of universalparticular is similar to
typeinstance in other analysis methods. In an information analysis,
one should describe properties of a type of things, rather than a par-
ticular instance, unless some particulars are so nearly unique and so
important that they have to be described specially. In our case study,
the departments, such as Land Tax, Valuation of Agricultural Land,
etc., are particulars, but they have to be treated specially in the
model.
Genericspecic Many phenomena in a social world may fall into a
genericspecic structure. The specics then inherit properties of the
generics. For example, in our case, the land parcel is a generic concept,
while the agricultural parcel and urban parcel are specics of the
generic concept parcel. The same kind of structure applies to various
departments.
Wholepart Some phenomena only exist as parts of a whole, through
which we come to know them. This is the wholepart relationship. For
example, in the case of an administrative structure, a district is a part
of a province and a village is a part of a district. Similarly, a street
exists as part of a settlement but not independently. However, in the
case of geographical areas, the parts pre-exist the administrative areas
to which they are allocated and, in this case, we do not have the
wholepart relationship but simply an assembly of components.
Role-carrierrole-name Some agents may be involved in certain actions,
where the agents are the role-carriers and have certain role-names. For
example, a person may own one or more land parcels, and then he is an
owner of the parcel. The relationship between the person and the
owned parcel is called ownership.
After having classied the semantic units, the following list is produced:
administration universal agent generic for province etc
province universal agent whole for district
district universal agent part of prov., whole for
village
village universal agent part of district
person universal agent
owner role-name person & land ownership
ownership universal aordance
land-unit universal aordance
parcel universal aordance generic for agri. & urban
parcels
agricultural parcel universal aordance a kind of parcel
urban parcel universal aordance a kind of parcel
tax-rate universal determiner
tax-bill universal aordance
Department of Valuation particular agent
of Agricultural Land
Safety and Emergency particular agent
Department
Cadastral Survey particular agent
Department
Department of Land Tax particular agent
. . .
Identify ontological dependencies
The ontological dependencies between phenomena are essential conceptu-
ally in Semantic Analysis. If one thing y exists only while x does, then the
dependence between them is dened as an ontological dependency. In
NORMA, it is denoted by
x y
indicating y ontologically depends on x. Note the relative positions of x and
y: the right item is a dependant, whereas the left item is an antecedent.
Using NORMA syntax, the ontological relationships between the
semantic units can be modelled as many fragments, as illustrated in Figure
12.1.
The process of constructing these fragments is actually to put each term
in its context with respect to ontological constraints. In this way each term
has to be assigned a clear meaning. The meanings of the terms are not only
clear to the analyst, but also clear to the problem-owner. If there is any mis-
understanding of the meaning of a term, it can be easily identied by the
problem-owner. Therefore both the analyst and the problem-owner can
cooperatively improve the problem denition or clarify the semantics of the
terms used in dening the problem through more interactions.
Complete the semantic model
The activity in this step is to assemble the fragments of knowledge into a
complete picture. The complete picture, a semantic information model,
describes the users requirements in a formal NORMA syntax, which in
itself is a design of the database, and can be directly put into implementa-
tion (see the section on design and implementation). The semantic informa-
tion model of Snake Island is presented in Figure 12.2.
The model is not claimed to be complete in covering all requirements
given by the problem denition. Some parts are not detailed enough, e.g.,
person
parcel
Dept.
Land Tax
parcel
tax-bill
owner
owner
ownership
liable to
prepares
tax
#amount
sent-to
tax-bill describes
the tax liability
Figure 12.1. Ontological dependencies in fragments.
land unit properties, information about parcels. They can be easily added to
the model. However, a substantial part of the requirements is covered in
this model.
To guide readers to read the essential parts of the model, an explanation
of some parts of the model is given as follows.
A district is a part of a province; a village is a part of a district (indicated
by a line with a dot attached).
All the four specic departments belong to the generic type of depart-
ment; but a specic department of Land Tax prepares the tax bills,
and another one, Agricultural Land Valuation, determines the tax-
rates of land parcels.
owner
@
depart.
Land Tax
Cadast. Survey
Agri. Land Valuat.
Safety & Emerg.
sent to
liable to
#amount
liability
province
district
village
has person
ownership
tax-rate
parcel
agri-parcel
urban-parcel
crop
building
on
land unit
planted
with
climate
soil
#temp
has
#type
#depth
tax
notification
@
Figure 12.2. The semantic model of Snake Island.
A person, as an owner, may have ownership of a parcel which may
specically be an agricultural parcel or an urban parcel. An agricul-
tural parcel (but not an urban parcel) may have several land units; a
land unit may be planted with many kinds of crops, and so on.
The aordance liability is a sign representing the legal liability and, of
course, a tax-bill is just a sign to notify a person of his liability.
Check the model
This is a task for both the analyst and the user. The analyst should present
the model and explain what is meant by the model to the user. Very often,
the semantic problems may still be discovered in this step, therefore this step
serves as not only a simple logic check, but also a continuation of the
knowledge elicitation and semantic clarication. Be prepared for the possi-
bility that this step may lead to a modication of some results of previous
steps.
12.3 Norm Analysis
The semantic model describes the agents and their possible actions in terms
of ontological dependencies, which are a kind of fundamental relation-
ships. Norms, in addition to the knowledge represented in the semantic
model, specify the details of these possibilities of behaviours; e.g., the con-
ditions where the actions must happen or where they are actually impos-
sible. The norms are less stable than the ontological relationships because
they are determined less by the underlying culture established in the com-
munity or organisation, and more by organisational expediency.
Types of norms
Norms can be categorised in a variety of ways (Stamper 1980). We distin-
guish, for example, between action norms, which direct or permit action,
and structure norms, which dene the hierarchical norm structure and say
when the dierent branches should be applied.
The action norms subdivide into ve main types which control very
dierent kinds of actions:
standing orders change the physical world
status norms change the social world
powers of intervention invoke or inhibit the use of existing norms
powers of legislation change the norm structure itself
structure norms dene the relationship between norms
Standing order Standing orders state the conditions under which some
concrete action should, should not or may take place. A computer can
perform the action if it has the appropriate electro-mechanical
devices but, normally, a person will have to do the job. In the case of
sending messages we can use the computer to implement the standing
order.
Example norm 1:
The owner or principal owner of a parcel of land should be sent a bill
for the tax on that parcel at least four weeks before the tax becomes
due.
Such a norm can be almost totally automated by the selection of rel-
evant people, parcels, and tax liabilities and by reference to the clock.
The bill is just a document that contains these details with a request for
payment. Someone will probably have to put the bills into the post
when the computer has printed them! The action involved is semiolog-
ical and computers are primarily devices for processing signs.
Status norms Status norms dene the legal and social status of the agents
and the actions. These norms give the conditions in which certain
kinds of legal status exist. Examples for this kind of norms are given
below.
Example norm 2:
If a parcel is owned by more than one person, the owner having the
largest share shall be the principal owner, otherwise any one of them
may be so designated.
This norm denes the status of a person to whom the tax-bill should
be sent in cases where there might be confusion. Nothing happens
until the status constructed is employed in a standing order.
Example norm 3:
If a parcel is owned by more than one person, each one shall be liable
to pay the whole of the tax on the parcel until full payment has been
received by the relevant authority.
This creates a number of liabilities, which will normally be dis-
charged by the principal owner collecting the shares of the tax and
paying the bill. It creates a social structure that will simplify the admin-
istration and generate pressures among the relevant group to solve
their problems of jointly administering their shared ownership in
whatever way they choose. The formal system involves no concrete
action.
Powers of intervention Powers of intervention allow parts of the law to be
brought into action or taken out of operation. They are like sub-
stations on an electrical power-distribution grid, enabling parts of the
network to be switched o and switched on.
Example norm 4:
Agricultural parcels which are in erosion or ood hazard areas, upon
application by the owner, may be exempted from land tax at the discre-
tion of the District Land Tax Inspector.
This norm creates discretionary power to be exercised by a person in
a designated role. A land parcel may be exempted from being taxed,
though the principle remains that every parcel should be taxed.
Example norm 5:
A claim for exemption from land tax under norm 4, if not allowed,
may be submitted on appeal to the District Land Tribunal.
This safeguards against a corrupt tax ocial by creating a supervi-
sory power of intervention.
Powers of legislation The powers of legislation dene the powers of
changing or creating laws and legislative procedures.
Example norm 6:
A District Land Tribunal shall specify locally appropriate procedures
for the conduct of appeals concerning exemptions from tax liability.
They have to perform an action of making norms, an action that is
several steps removed from concrete acts for which standing orders
will generate the necessary commands.
Example norm 7:
The Minister of Land Resources shall inspect the appeals procedures
of the various District Land Tribunals and issue any necessary regula-
tions concerning the conduct of appeals in order to maintain adminis-
trative eciency.
Here the minister is empowered to make norms within constraints
that, in principle, the courts could impose if he made norms that did
not help to maintain eciency.
Structure norms are the norms about norms. The structure norms consist
of a condition clause determining when a set of norms should be
applied. Unlike powers of intervention, structure norms are categori-
cal and do not introduce discretion; they function simply to represent
a logical feature of the norm hierarchy.
Example norm 8:
For the purposes of assessing liability to land tax, the extent and
location of erosion or ood hazard areas shall be determined in
accordance with the provisions of the Agricultural Development
(Marginal Areas) Act 1976.
Locus of structure semantic model versus norm structure
Many conventional information systems built upon the foundation of a
relational database prove to be uncomfortably rigid in practice, when faced
by organisational changes. One of the commonest reasons for this is the
tendency to incorporate into the database schema features that are rela-
tively supercial. These exible features are norms rather than semantic
structures. For example, Example norm 3 above establishes, in eect, that a
liability to land tax cannot be subdivided into parts but it is not inconceiv-
able that the law could be changed in order to allow this. The problem for
the designer is whether to treat the individuality as a feature for the schema
or as one to be handled by the application programmes. Decisions of this
kind are not always easy but the preferred solution is always to treat con-
straints that are not ontologically necessary as being dened by norms. In
the Snake Island problem we are studying here there is a case in point.
Example norm 9:
No building shall be erected on an agricultural parcel of land.
This may be incorporated into the model by giving a building a relation on
that is conned to urban parcels, and not to the other kinds of parcels.
Identication of norms
Sometimes we can build computer systems to support the administration of
existing legislation, codes or regulations. This gives us an obvious way of
identifying the norms that must be built into the system. However, it does
not guarantee that the norm system will be complete. Here the semantic
model gives us valuable support as a checklist. Once the semantic model is
obtained, the boundary of the requirements is clear but, also, the items
specied in the semantic model identify all the norms required for their reg-
ulation. This is because the norms are the elements of the system
specication that dene the dynamics and we can associate with each
element in the model the start and nish events. Every one of these events
must have its (formal or informal) governing norm. These norms can be as
complex as the problem domain requires.
12.4 System design and implementation in the Normbase approach
The Normbase system can assist one to achieve the following:
accepting a semantic model as an input,
automatically turning the model into a time-based database design,
helping users to capture data with a user-friendly interface and consis-
tency checking,
oering users standard application tools, e.g., query and report,
simplifying system maintenance.
12.4.1 From semantic model to database design
In the Normbase system, there are predened meta-classes. Each meta-
class has a standard data structure. As soon as the semantic model is input
into the Normbase system, all the agents and aordances have their
predened data types. The ontological dependency relationships in the
model are also captured in the system. The ontological dependencies are
established between particulars as well as between universals. For example,
in the land resources information system, there are universal relationships
between the universals person and parcel. And, at the particular level,
there are particular persons and particular parcels between which there are
particular ownerships.
STDB
The STDB as a kernel part of the Normbase system handles data in a struc-
ture that incorporates periods of existence (start time and nish time) as
their attributes. For data manipulation purposes, a time-based language,
LEGOL, is employed. It can be used for two purposes: to write queries and
to write norms.
Queries
Most LEGOL operators are time related, for example, the operators while,
or-while, and while-not. When these operators are used, they perform oper-
ations in two aspects, i.e. set operation and time operation. The following
examples will illustrate how the LEGOL operators work in performing
queries on the database.
Based on the semantic model in Figure 12.2, suppose the user has the fol-
lowing questions:
Question 1: Who owns at least both one agricultural parcel and one
urban parcel at the same time?
Question 2: Which parcels did/do not have buildings on?
Question 3: Which urban parcel had the highest tax-rate last year? And
any time up till this year?
Question 4: Has anyone been notied of their tax liability more than
three months after becoming liable?
12.4 System design and implementation in Normbase 175
These questions can be translated into queries in LEGOL as follows.
Query for the question 1:
ownership(person, agri-parcel) while ownership(person, urban-parcel)
Or using role-names in the query,
person while owner(agri-parcel) while owner(urban-parcel).
These two queries will return the following information:
the persons who own at least one agricultural parcel and at least one
urban parcel at the same time;
the particular agricultural parcels and urban parcels those persons own;
and
the periods of these ownerships over the parcels.
urban-parcel while-not on(building, urban-parcel)
This query will tell which urban parcels, in which time periods, did/do
not have buildings on them.
The query below, however, can generate the answer to a question like
Which urban parcels have never had buildings on?:
urban-parcel where-never on(building, urban-parcel)
highest tax-rate(parcel) while last year
The last query will tell us the parcels with the highest tax-rate last year.
But the next one,
highest tax-rate(parcel),
will tell us the parcels with the highest tax-rate over the whole recorded
history of the tax regime.
Notice, in the semantic model in Figure 12.2, there are hash signs before
amount, type, depth and temp. to indicate they are determiners (i.e., a gener-
alisation of the measurement concept). With a numerical determiner, it is
possible to use the functions like highest, max, etc.
Query for question 4:
start-of sent-to(owner, notication(liability)) after (3 month start-
of(liability)
In the LEGOL expression, start-of is an operator which gets the start time
of its operand, i.e. sent-to, and liability. Also after is an operator which
produces a valid time period after a given time point, and then performs a
joint operation on its left and right operands.
Norms
Norms can be written in LEGOL. These norms are then stored in the
Normbase system and they are linked to the relevant agents and
aordances in the database. The linked norms can serve two purposes:
As constraints to check operations on the data items. For example,
Example norm 4 listed earlier should be linked to the agricultural
parcel. When the tax-rate is determined on the parcels, that norm 4 has
to be evaluated to see if the parcel has an exemption from tax or an
outstanding application for exemption.
As triggering norms to prompt actions. These actions may be of two
kinds. One is automatic change of data in the database, which is done
by the norm triggering mechanism within the Normbase system. For
example, if there is a tax-calculation norm written in LEGOL attached
to the tax-rate, the tax-rate can then be calculated automatically
according to the updated information about the parcel, which is kept
in the system. The other kind of triggering action is to produce mes-
sages to users, prompting them to perform the necessary action in the
business world.
Generation of a STDB
In the Normbase system, the records that keep information about agents
and aordances are called surrogates. All the surrogates have a uniform
data structure. Below there is a denition of the surrogate as it is imple-
mented in the existing version of the system.
surrogate :surrogate-identity, type, sort, label, antecedent-1,
antecedent-2, start-time, nish-time
Keeping historical data
Sometimes to keep historical data in databases has signicance, especially
in the application domains where legal responsibility is much concerned.
These are domains like banking, contracting, government aairs, as well as
land resources management. The Normbase system oers a kind of data-
base able to keep all the historical data. This is realised as in the example
demonstrated below.
Suppose in the database there are three surrogates as follows, in which x1,
x2 and x3 are values of the tax-rates for land parcel p0:
12.4 System design and implementation in Normbase 177
201, particular, tax-rate, x1, parcel#p0, 1-7-1988, 30-6-1989
These surrogates will give historical information on the tax-rate for the land
parcel p0 over the last few years.
Non-destructive update
The Normbase system adopts a non-destructive way of updating. Two
kinds of time are recognised and treated in the Normbase system, the event
time and the transaction time. The event time is to record the existence of a
happening in a real world, such as the start time and nish time used in the
surrogates. The transaction time is to indicate when a surrogate is inserted
into the database and when it is updated. But the surrogate to be updated
will not be removed from the database.
The Normbase system realises this non-destructive updating by attach-
ing a transaction time tuple to each surrogate, as illustrated below.
data-record :surrogate, transaction-time , where
transaction-time :transaction-start, transaction-nish.
Suppose there are the following data records in the database about a land
parcel p0.
1-7-1988,
1-7-1989, 2-7-1989
2-7-1989,
1-7-1990,
Notice the second and the third data records are almost the same, but with
dierent values of tax-rate and transaction time. The reason for this may be
that the tax-rate for the parcel p0 for the period 1-7-1989 till 30-6-1990 was
rst entered into the database on 1-7-1989, but it was discovered it should
be x3 and corrected on 2-7-1989. So the valid data records have a null value
for transaction-nish time; the data records with lled values of transac-
tion-nish time are not applicable, and possibly there are updated data
records for them.
12.5 Discussions and conclusions
Using Semantic Analysis, much more attention is required for the analysis
and requirement specication than in the other approaches. One has to
follow the analysis formalism to undergo a rigorous process in order to
produce a stable model. After having obtained the semantic model, the
Normbase system tools can turn the model directly into a database imple-
mentation. Meanwhile there will be standard application tools provided.
Therefore the emphasis is shifted to the analysis stage from a users point of
view. Because not so much designing and programming is required, the cost
of system development can be dramatically reduced.
Semantic Analysis oers a powerful analytical method for information
analysis. Together with the Normbase system, it suggests a new way of
information system development. This new approach demonstrates how to
capture application semantics in an information model, and suggests a
complete solution for system development which is more exible and eco-
nomical in meeting users requirements. The time-based DBMS, the time-
based database language and the non-destructive updating method
adopted in the Normbase system also support, in a practical way, the man-
agement of information in the legally oriented areas where it provides a vir-
tually perfect audit trail on account of the non-destructive update.
Finally, it should be noted that in countries that are rich in competent
people but short of foreign exchange to buy costly technology, these
methods allow careful analytical thought to be substituted for redundant
technology.
12.5 Discussions and conclusions 179
13
Case study: development of a test
construction system
This case study demonstrates howsemiotic methods canbe usedfor systems
analysis, and the other stages of systems development are covered by other
approaches. The chapter also discusses why, after having tried other systems
analysis methods, the Semantic Analysis methodwas chosenfor the project.
13.1 Background
CONTEST (COMputerised TEST construction system) is a software engi-
neering project which started from research and then evolved into a com-
mercial project.
1
In this project, Semantic Analysis has been applied for
requirement analysis. An object-oriented system design has been produced
based on the semantic model, and, further, the system construction has
been carried out using object-oriented programming languages and tools.
The project commenced from two doctoral research projects, in which
the major theoretical investigation and some experiments were conducted
(Adema 1990, Boekkooi-Timminga 1989). A system analysis and design
were performed which led to a prototype. The research project has been
continued and expanded with input of more resources. The current objec-
tive of the project is to produce a practical, useful, computerised system for
automatic management and production of tests.
13.1.1 CONTEST project
Tests are traditionally managed and produced manually. The questions (or
the items as termed by the educational professionals) are written by the item
180
1
CONTESTwas a joint project of the University of Twente, the University of Groningen and
the Inter-university Expertise Centre ProGAMMAin the Netherlands. The development of
the rst version of CONTESTwas completed in mid 1994, which is the basis for this chapter.
composers, and centrally collected and stored with their attributes in an
item bank in a test agency. These test agencies are specialised in managing
and providing tests for all kinds of examinations. Examples of such agen-
cies are the Educational Testing Service (ETS) and American College
Testing program (ACT) in the USA, and the National Institute for
Educational Measurement (CITO) in the Netherlands. Nowadays several
agencies are building item banks in the form of computerised databases.
However, in most of the cases itemindexes are still used for assembling tests.
The goal of CONTEST is to function as an interactive system for the
storage of items, the construction of tests, and the editing and printing of
tests. Test construction is the process of selecting items for a test from an
item bank in such a way that the test ts the specied requirements (i.e., test
specication) formulated by the test constructor. A test specication con-
sists of two parts: an objective function to be optimised and a series of con-
straints. The test specication is formulated in relation to the examination
purposes. CONTEST aims to aid the users formulating specications by
building a knowledge base from which the expertise and experience of
making proper specications can be made available. With reference to the
test specication, tentative selections of items must be checked against the
specied conditions, and modications on these tentative test proposals
will be adjusted until the conditions posed in the test specication are fully
satised. Afterwards, the test will be printed for use in examinations and
educational quality analysis and for other purposes.
13.1.2 User requirements
The beginning of 1990 can be seen as the beginning of the project in which
the computerised test construction system CONTEST became the objec-
tive. The knowledge gained in previous years had laid a foundation from
which more sophisticated requirements on CONTEST were derived. The
following system components and their functions were formulated as pre-
liminary requirements, though some of them were not fully articulated.
An item bank Seven dierent data sets are considered in CONTEST:
persons, items, tests, graphics, auxiliary materials, ability scale, and
populations. Besides the contents of the items, characteristics of each
item have to be described. In Table 13.1 an overview is given of some
possible item characteristics that are stored for each item.
Information about the uses of an item must be kept in the item
bank. Information on the tests already constructed from an item set
includes, for example, test specication used, items selected, date of
13.1 Background 181
test administration, and group of students to which the test was
administered. The graphical parts of the items are stored in the set of
graphics. In the set of auxiliary materials information on auxiliary
material belonging to (some of) the items is stored. For example, for a
reading comprehension item with text passages as auxiliary material,
for instance the length of the text passage and/or its required reading
time can be stored for each item of auxiliary material. Furthermore,
information on possible student populations or abilities for which tests
can be constructed is stored.
182 Case study: test construction system
Table 13.1. Item attributes.
1. General attributes:
Item ID
Item description
Item text
Length of item text
Word count
Item graphic ID
Item answer
Item format
Item type
Author of the item
Date the item was written
Gender orientation of the item (male/female/neutral)
Ethnic orientation of the item (black/white/Hispanic)
Response time
2. Subject matter attributes:
Content descriptions
Cognitive level
Relation to textbook(s)
Auxiliary material ID
3. Psychometric attributes:
Classical item parameters:
p-value ( item diculty)
Item test correlation ( item discrimination)
Item criterion correlation
Item reliability
Item validity
Item response model (Rasch, 2- or 3-parameter model)
Diculty
Item discrimination
Item t
Dierential item functioning
Item guessing
A test construction mechanism An essential component of CONTEST is
the mechanism for constructing tests. A new test specication is for-
mulated, or an existing test specication can be loaded from the test
specication storage and adapted if desired. The test specication con-
tains the practical requirements for the test. With the help of mathe-
matical optimisation techniques (e.g., linear programming), items are
selected such that all requirements are met. If no solution is found, the
system can check which requirements are likely to cause the problem.
Otherwise the characteristics of the test and the items selected are
shown to the test constructor. It is possible to compare the characteris-
tics of this test with other existing tests. The test selected, its
specication, and its attributes are stored in the system. Finally, the
test layout is edited and the test is printed together with its answer
sheet.
A mathematical optimisation module In the mathematical optimisation
module, an objective function maximises or minimises the goal value
subject to a number of constraints. For example, a mathematical opti-
misation model can be to minimise the test administration time subject
to the constraint that at least 40 multiple choice items be included in
the test. Linear programming models can be used to solve specic
types of test construction problems which assume that both the objec-
tive function and the constraints are linear expressions in the decision
variables.
A package for handling test specications This module simply must have
functions of storing all the test specications that have been used for
constructing tests. These specications may be useful references for
later applications.
A knowledge base assisting elicitation of test specications The users of
CONTESTdonot have tobe familiar withthe mathematical models. A
knowledge base which supports the elicitation of the test specication
should be provided. The knowledge base contains the knowledge
needed for eliciting the test specication, such that the corresponding
linear programming model can be derived without involving mathe-
matical notation. The use of this knowledge base makes it possible to
elicit test specications from a test constructor for all kinds of item
subsets with their own specic structures. The test specication
depends on the item characteristics stored in the specic item subset
and the desires of the test constructor. Both inuence the possible
objective functions andconstraints. Besides the assistance for choosing
an objective function and constraints, the knowledge base supports
13.1 Background 183
formulating test specications for constructing special types of tests,
for example parallel tests. For each special type of test to be con-
structed, knowledge is stored in the knowledge base that helps the
choice of objective function and constraints.
13.1.3 Why choose Semantic Analysis
The reasons for choosing Semantic Analysis came from the special require-
ments of the project, summarised as several challenges to the conventional
methods for analysis, design and implementation.
The challenge can rst be attributed to the fact that there are many stake-
holders in dierent states of the system development lifecycles. The project
is conducted by a multidisciplinary team. The researchers are the people
who have knowledge about the subject matter of the system. They have
been generating ideas and producing requirements. However, they are not
the real users of the system. The ultimate real users are test agencies and
educational institutions. There are several groups of users working with the
system. Item composers produce items and store them in the item banks.
Test constructors specify the kind of tests to be constructed out of the items
available in the item bank, by detailing the purposes of the test and all kinds
of constraints. The test constructors then produce test proposals and
nalise the test. During the system development, the educational
researchers in the project team have played a leading role in deciding the
functional requirements. In addition to these knowledgeable people, there
are systems analysts, designers and programmers who have been involved at
dierent stages in dierent tasks, though some of them played more than
one role in the project. The complication caused by this situation is that
people from various backgrounds have seen the problem from dierent
angles and also presented it dierently. Jargons and specialised terms were
used in interdisciplinary communications which created many misunder-
standings. Many crucial terms for the description of the subject needed to
be claried. A question for the project team was how to use words and
terms to formulate requirement specication. To state the question more
directly: whether new languages which may be suitable for system develop-
ment (particularly analysis, design and implementation, etc.) should be
created, or at least some new terms, because one does not understand the
technical dialects of others. For example, the concept item is crucial in the
CONTEST system; the educational professionals speak about an item for
a question to be used in a test or examination, which was a surprise to the
system development people. On the other hand, the documentation of
requirement modelling and system design was dicult for the educational
experts to judge whether the representations correctly reected their wishes
and desires.
The second challenge came from the ambitious and complex require-
ments. In the early phase of the project, there was a desire to develop a com-
prehensive system based on the available theoretical work and experiences.
However, diculties were encountered when the desire had to be expressed
as a formally articulated requirement. In addition, the original requirement
possibly continued to evolve as the educational researchers had some new
inspirations or some new theoretical discoveries.
The third diculty lies in the technicality. The data objects to be handled
in the item bank will include simple data types, mathematical formulae, text
and graphs. In the test specication part, specications must be stored for
re-use and knowledge for formulation of specications must be kept avail-
able. In the test construction process, there will be mathematical optimisa-
tion and human interactive intervention involved. In summary, what has
been required is actually a comprehensive system which comprises a knowl-
edge-based and an object-oriented database as well as a mathematical opti-
misation mechanism. The systems overall architecture, as illustrated in
Figure 13.1, was gradually envisioned during the project (Boekkooi-
Timminga & Sun 1991). However, according to the study of the market,
13.1 Background 185
Graphic user system interface
Editor
Report generator
Specification
editor
KBMS
Knowledge
base
Test
constraint
formation
mathematical
optimisation
DBMS
database
Figure 13.1. The CONTESTsystemarchitecture (Boekkooi-Timminga&Sun1991).
there was no software package or database system that could be adapted for
these purposes.
At an early stage of the CONTEST project, ISAC (Lundeberg et al.
1981) as well as the entityrelationship method (Chen 1976, Howe 1989)
was applied for systems analysis. But the project could not continue, for
several reasons. The methods do not put emphasis on the semantic problem
of concepts and terms, which is typically required in a multidisciplinary
project such as CONTEST where some terms are used with several dierent
meanings. The second problem that occurred was that the results of the
analysis were dicult to comprehend by the team members except for the
analysts themselves. The large volume of documentation created a barrier
to comprehension so that the colleagues from the educational background
were not able to conrm whether the analysis met their requirements.
However, the project went into the design phase with these problems
unsolved. In design the entityrelationship method was applied, and it was
planned to use a relational database package for implementation. It was
also discovered that the use of a standard relational database was not suited
for CONTEST for several reasons.
(1) At that time standard relational database packages could not store
graphs.
(2) The structure of the items to be stored is not xed.
(3) The lengths of the data cannot always be xed (e.g. the item texts can
vary very much in length), resulting in memory problems if a relational
database is used.
(4) Eliciting and storing test specications would be very problematic,
because it would be very dicult to store all possible constraints and
objective functions for each characteristic of an item in the database for
an arbitrary test.
After all these trials, Semantic Analysis was applied. Semantic Analysis
immediately seemed to be suitable because it represented the knowledge of
the universe of discourse with clear meanings; the model is easy for the pro-
fessional colleagues to comprehend and they were then able to criticise and
make improvements on the model.
13.2 System analysis
The method of Semantic Analysis has been chosen and applied to the
CONTEST system. Two major benets have been experienced from the
application of the method of Semantic Analysis. The rst one is that a
principle of Semantic Analysis is to require analysts to adopt the users
language. The creation of new words or names only for representation
purposes is virtually forbidden. Even the use of abbreviations in the
semantic chart is not recommended. This will reduce not only the lan-
guage distance between the problem-owner (or project-owner, in this par-
ticular case) and the analysts, but also the psychological barriers. The
problem-owners do not have to use a data dictionary to read the require-
ment documentation as long as the rules of representation are explained.
However, there is a cost involved. The analysts must make an eort to
understand the problem and to learn to speak the problem-owners lan-
guage. They are discouraged from inventing names and labels in represen-
tation. No ambiguity in documentation is allowed to hide behind invented
jargon and technical terms.
The second benet from adoption of Semantic Analysis is the minimal
volume of the documentation. Previously the large quantity of documenta-
tion of analysis was a barrier for the educational experts examining the
result of analysis produced by other methods. With Semantic Analysis, the
major documents produced from analysis were three pages of ontology
charts. One leading educational researcher has actually been able to take
part in the construction of the semantic model. Other educational profes-
sionals were able to appreciate the complete semantic model because it is
clear and concise. Therefore, there have been iterative and critical examina-
tions of the result of Semantic Analysis by the problem-owners. Finally, an
approval by the problem-owners of the system requirement model was
given on the basis of a full understanding of the analysis, which oers a reli-
able ground for further steps in the system development.
The system analysis was carried out in the following phases.
1 Study problem description
A description of the problem was obtained during several discussions
within the project team. The description was circulated to a wider circle of
people concerned with the system in one way or another for criticisms and
suggestions. Furthermore, theoretical work was studied and existing test
construction systems were evaluated. In this phase, the task of the whole
team was to formulate the user requirements at a global level. For the ana-
lysts it was, at this phase, the rst opportunity to learn the problem, there-
fore they had to grasp the concepts and terminology.
The user requirement formulated in the problem description was not
detailed and precise, but it was not necessary for it to be so. The next phases
would provide chances for improvements.
2 Identify semantic units
In this phase the semantic units were identied. The focus at this stage was
on the identication of the agents and aordances. The semantic units
contain specic meanings in terms of describing the agents and the possi-
ble actions in the problem domain. In other words, they are candidates
which may be put into the information model to represent the perceived
object world. Some of the semantic units identied were listed as in Table
13.2.
While identifying these units no attention was paid to the proper group-
ing of them. Some of the terms were just listed, without knowing whether
they would be useful for describing the problem. However, all the terms
which seemed to be relevant in dening the problem should be listed as the
semantic units. They provided a basis for a formal requirement formulation
in the next two steps.
3 Group semantic units
In this phase the semantic units were grouped: the agents and their
aordances, and the ontological antecedents and their dependants. The
agents and their aordances were related in such a way that ontological
consistency was obtained. For example, an agent person has an aordance
write an item; a piece of knowledge a group of students from a certain
school are administered a test; and a test is composed of certain items are
represented as in Figure 13.2. Several such scattered groups of semantic
units were the result of this phase, which are to be assembled into a com-
plete ontology model in the next phase.
Table 13.2. Some candidate aordances.
person
writes [an item or test question]
writes [a test specication]
writer
institute
school
item [a test question]
gender
response time
ability scale
population
test
composed-of [a test] . . . . . . . . .
4 Construct the semantic model
In the nal phase a complete knowledge model, or ontology model, was
derived by collecting all the groups of semantic units obtained in Phase 3.
This was not a simple matter of assembling the pieces of knowledge, but a
process of putting fragments of knowledge into a large semantic context in
which all the pieces must be ontologically and semantically consistent.
Critical evaluation of the model during and after completion is very
important. The problem denition was carefully studied again and the
ontology model was cautiously checked. During the whole analysis the pro-
fessionals and analysts carried on the review continuously for some time. In
fact the evaluation was a continuing process of knowledge acquisition.
The professionals often use specialised terminology that they perhaps
mistakenly expect others to understand. Such terms need to be claried;
otherwise, serious problems can arise later and a system can be built that
does not t the requirements of the problem-owners. Semantic Analysis
produces the ontology chart as a platform for the professionals and ana-
lysts to reach an agreement on the terminology used; it does not allow the
analyst to make arbitrary assumptions about the meanings of words.
Leaving out the many determiners that describe the attributes of the
aordances attached, a major part of the semantic model obtained is pre-
sented in Figure 13.3. In the gure a line along with a dot represents a
partwhole relation, for example, the partwhole notation shows that an
item can contain a graph and a text. A person can write items and write a
test specication; in both cases they have a role name writer but are
school
from
student-group
administered
test
person
write
item
test
item
contains
Figure 13.2. Grouping semantic units.
involved in dierent actions. The action of selection is actually a complex
process including the generating of a constraint matrix, optimally selecting
items according to the constraints, etc. For simplicity of illustration, the
details are not shown on the gure. The automatic selection process may
lead to the existence of a test if the process is satisfactory according to the
test constructors judgement; therefore, a dotted line along with an at-sign
(authority sign in NORMA) is used to indicate that the selection deter-
mines the start for the test. After the test is constructed, it is reviewed,
edited and printed. The nal test is administered to a specic student group.
Once the semantic model has been obtained and approved by the
problem-owners, it serves as a formal description of system requirements.
Moreover, it has been a basic documentation for system design which will
be discussed in the next section.
13.3 System design
The semantic model preserves ontological dependencies between agents
and actions. Ontological relationships between agents and actions
described in the ontology chart reveal that there are usage (e.g.
client/server) relationships between them. Actions that are aorded by
e
m
p
l
o
y
e
r
e
m
p
l
o
y
e
e
writer
@
person
root agent
student group
administered to
works for
writes
test specification
selection
writes
item text test
contains
graph
school
test institute
institute
w
r
i
t
e
r
Figure 13.3. An excerpt from the semantic model of the CONTEST system.
agents suggest functions to be performed by objects in the object-oriented
paradigm. For example, in Figure 13.3, between the item, test-
specication and test there are ontological relationships and actions (e.g.
the action of selection of items by a mathematical optimisation procedure).
An important task at this stage is to transform the relationships and actions
in the ontology chart to the form of system design.
The object-oriented approach has been chosen for the design and con-
struction of the CONTEST system. The rationale for the choice is that rst
of all the Semantic Analysis itself lends the result of analysis readily to the
object-oriented design. There are principles for transformation from a
semantic model to an object-oriented design. Identications of objects,
attributes and functions of the object, and relationships between the
objects are obtainable from the semantic model. The second reason for
using this design method is that the path from the object-oriented design to
the object-oriented programming is smoother than if other design methods
were used. These implementation decisions, based on the feasibility study
of the software tools and packages on the market, were taken at the same
time as the design method was chosen.
The designprocess canbe presentedinfour aspects: class diagrams, object
specications, object lifecycle diagrams, andclass inheritance diagrams.
As an example, the part of the semantic model containing the four
aordances item, test-specication, test, and selection, is used to show
the transformation into a class diagram (Figure 13.4). Three objects are
obtained after the transformation with the same names as they have on the
ontology chart. The aordance selection is transformed into the commu-
nication action between the objects because it is an action-like aordance.
This part of the design also shows that one object uses another in certain
ways. The dependency relationships are treated as attributes, while the
actions in the semantic model are turned into functions and procedures in
Item
TestSpecification
Test
1 1
n
n n
n
Figure 13.4. Part of the CONTEST class diagram.
object-oriented Pascal (or called member functions in C). A cloud repre-
sents a class. The class has a name which is placed inside the cloud. The
usage relationships are the most important kind between classes. They are
indicated by the double lines with circles placed at one end. For example,
the double line between the classes of Item and TestSpecication is used
and then a lled circle is placed near the class of the TestSpecication,
asserting that the class of TestSpecication uses the resources of the class
of Item in its implementation. A lled circle and an unlled circle are
placed close to the class of Test, asserting that the class of Test uses the
resources of the class of Item in its implementation and interface as well.
The notation for cardinality among the objects is used. As shown in the
gure, for every test object, there is exactly one instance of the class
TestSpecication and for every instance of the TestSpecication, there is
also exactly one Test. There may be many instances of the class of Item
associated with each test, also many tests for every item.
Object specications are formal descriptions of the objects. In the
CONTEST project, a special structure was invented for documenting the
composition of the object. Each structure consists of three parts: the
Table 13.3. Illustration of object descriptions.
Test specication Test Item
attribute: attribute: attributes:
TestSpec_ID:number; Test_ID: number; Item_ID:number;
Test_ID:number; TestName:character; Description:text;
ConstraintSet: set; Description: text; AuxiliaryMaterial:
Operator: string; DateOfConstruction: composite;
Bound: real; date; DateOfCreate: date;
function: function: Content: set;
initialise; add; update; initialise; add; update; CognitiveLevel:set;
delete; test_construct; TextBook: booktitle;
view_constraint; matrix_input; Author: name;
Diculty: real;
Guessing: real;

function:
initialise; add;
update; delete;
View_on_text;
View_on_barchart;
select_on_characteristic;
count_item;

heading on which the name of the object is placed, the attributes, and the
functions (or actions that the object can perform). Table 13.3 presents
examples of object specications for three objects: TestSpecication,
Item, and Test.
The lifecycle of objects is described with state transition diagrams.
Figure 13.5 illustrates the state transitions of two objects:
TestSpecication and Test. The object Test, for example, has two states:
test draft and test print. Below the state boxes, possible actions in those
states are listed, each preceded by a dot. These actions are the functions
that have appeared in the structure of the object specication. Events that
bring objects from one state to the other are placed along the transition
lines; for example, the events modify, elicit and convert change the
states of the object TestSpecication. During their lifecycles there are
interactions between objects which are caused by the external events. These
external events are placed in dotted boxes, e.g., execute math model and
change proposal. This kind of event sometimes synchronises states in the
lifecycles of both objects.
Another aspect of the design is the class inheritance, for which the class
inheritance diagrams were used. This aspect shows that the parent object
classes have common structures and properties, and the children classes can
have some added, specic functions and properties based on the parent
classes. For example, an institute, as a parent, has a name, can employ sta,
TestSpecification Test
change
proposal
1. specification
proposal
1. test draft
2. constraint
table
2. test print
modify
execute
math model
view items
assemble items
file
accept test
view
edit
file
elicit; convert
edit
print
normalisation
enter math model
file
Figure 13.5. Object state transitions.
etc. A test institute and a school are specic types of institute and can have
additional functions and properties unique to the specic classes of objects.
13.4 System construction
The CONTEST system was constructed using an object-oriented Borland
Pascal. In addition, other object-oriented utilities are employed, such as
Object Professional and B-Tree Filer. The former is a class library which
provides object implementations for screen, windows, menus, pick list, data
entry screen, and so on. The latter, B-Tree Filer, was used to build up the
object-oriented database management system for the large database with
complex data types, because there was no suitable object-oriented database
package available when the project was initiated. Several pre-release ver-
sions of the CONTEST system had been developed by the time of report;
in each of them, more functions were added. Positive reactions have been
received when the system was demonstrated to some potential users.
13.5 Discussion and conclusions
In the case study of the CONTEST project, in which a multidisciplinary
team is engaged to develop the information system for computer-assisted
education, an emphasis has been put on analysing and specifying user
requirements. Semantic Analysis and an object-oriented approach have
been chosen to cover the whole process of the system development, namely,
analysis, design and implementation.
Application of Semantic Analysis has brought several advantages to the
project. The method takes the users language as the starting point to
describe the agents and actions in practical aairs. First of all, the semantic
model is a result of articulation and clarication of meanings of the users
conceptualisation represented with a few formal constraints (e.g. ontologi-
cal dependency). This lends the model itself to be easily understood by the
project participants (specically, the educational professionals) who should
actually be in control of the project (or project-owner, in other words).
The second advantage of the use of Semantic Analysis is that the result
of analysis was presented in a precise and small volume of documentation,
which allows the educational professionals to have the courage to walk
through the documents with the analysts. Therefore, the project-owners
were able to understand, and criticise and make corrections to the model.
Another positive aspect is that the ontology chart served as a conceptual
basis for the object-oriented design. The principles of transformation from
a semantic model to an object-oriented design ensure the design quality to a
very large extent. Adopting rigid object-oriented disciplines also secures
the quality of the product as a whole. Having applied this set of approaches,
the whole team, both the project-owners and the system developers, feel
condent to commit themselves to put resources into the project. This has
been considered as one of the most signicant contributions of the
Semantic Analysis to such a large-scale, complex system development
project.
However, one unsatisfactory aspect was experienced in the project. In
application of the system development approach which is based on
Semantic Analysis and supplemented by the object-oriented technology,
the transformation from an ontology chart to an object-oriented design
was not straightforward. The transformation principle could not fully help
the exercise because the distinction between the agent/entity-like
aordances and the action-like aordances is not explicitly made in
Semantic Analysis in the rst place. Therefore, as compensation to the
problem, the designers had to learn Semantic Analysis and the ontology
chart. The ideal solution would be to provide a highly formalised
specication technique for representing the result of analysis with which
the design process can proceed without having to know the details of the
semantic modelling.
13.5 Discussion and conclusions 195
Appendix A Semantic templates and
surrogate specication
This appendix, as a supplement to section 10.2, supplies further details of Semantic
Templates (ST) and the use of ST for database design.
A.1 Denition of ST
The denition of ST, basically following a syntax of BackusNaur form, describes
the structure of a semantic template.
ST : agent, ST, mood, mood, time, time | ST : agent, action,
time, time;
agent :agent performing the action;
action :realisation of aordance;
mood:mood;
mood:mood;
mood : proposal | inducement | forecast | wish | palinode | contrition | assertion |
valuation;
proposal :request|ask|beg|command|insist|suggest|promise|guarantee;
inducement :reward|threaten|warn|tempt;
forecast :predict|assume|plan|accept;
wish :want|hope|expect|desire;
palinode :retract|annul|revoke|deny;
contrition :regret|apologise;
assertion :assert|claim|report|notify|arm|assure|argue|declare;
valuation :judge|appraise|blame|criticise|accuse|object;
time:time;
time:time;
time :absolute time or relative time (by reference to another action).
The right-hand phrases in italics are explanations of the possible value types of the
left-hand items. For example, the agent is dened as agent performing the action,
which refers to the agent appearing on the ontology chart and relating to the action
as the antecedent. The action is dened as realisation of aordance; the
aordance should also be predened on the ontology chart so that there can be a
realised instance of the aordance. This can be illustrated by relating to the example
shown in Figure 9.4, and it will be used throughout this appendix. In that gure, a
person, John, is the agent; a particular case of Johns being absent is a realisation of
the aordance absence.
196
The moods are categorised into eight headings. Under each of them, there are
some typical illocutionary words. This organisation oers an opportunity for
further analysis of the generalised communication patterns. For example, a projec-
tion may lead to an anticipation by the hearer; a directive will need a semiological or
substantive action by the hearer while an assertion may need no action at all (except
the hearer changes his belief, for example). This kind of categorial analysis can be
helpful for building an engine into a technical information system for automatic
assistance to communication.
A.2 Examples of using ST in discourse modelling
In this section, more examples of using ST in modelling knowledge expressed in lan-
guage will be given. An ontology chart resulting from a semantic analysis is a basis
for ST modelling. The illustration in this section can be seen as a continuation of
section 9.5, using the same case (shown in Figure 9.3).
The following passage tells six stories about Johns leave. Afterwards, these stories
will be represented in ST form, with an intent to capture semantics, intentions and
subjectivity of the discourses. Each of the STs represented will be supplied with
explanations.
Narrative 1: John was absent from 1-6-90 till 30-6-90.
Narrative 2: John started to take leave from 1-6-90 and planned to be back on 30-6-
90.
Narrative 3: Susan thinks (assumes) that John may be o from 1-6-90 till 30-6-90.
Narrative 4: Susan thought (assumed) 31-5-90 that John may be o from 1-6-90.
John planned on 15-5-90 to leave from 1-6-90 till 30-6-90. But actually
he left on 2-6-90 (discovered on 5-6-90). Finally it is observed that
John took leave from 2-6-90 till 20-6-90.
Narrative 5: Peter said today that yesterday Susan asked if John takes leave from 1-
6-90 till 30-6-90.
These ve stories will be described in the ST form in the following ve episodes.
Episode 1: , John, absence, course 123, 1-6-90, 30-6-90, assert, assert,
,
Remarks: (a) This rst item in the tuple is the agent who performed the speech
act. The agent in this case is not specied for some reason. The
agent may be the teacher of John or an ocer of student aairs
who reported this.
(b) The semantic content John, absence, course 123, 1-6-90, 30-6-
90 is determined by relating it to the corresponding part of the
ontology chart, which virtually species who John is, what is
meant by absence, and what the course 123 is.
(c) The start time and nish time 1-6-90, 30-6-90 represent the
period of the action Johns being absent.
(d) Both the start mood and the nish mood are assert, which means
the two dates for start and nish of Johns being absent are
asserted observations. In other words, the speech act represented
in this episode is an asserted (or reported, observed ) fact.
(e) The two dashes in the tuple represent the fact that the times of the
performing of the speech act are unknown. With an unknown
start time, this means the time when the speech act was performed
is not recorded (it might be irrelevant to the subject). The unlled
Semantic templates and surrogate specication 197
nish time indicates that the speech act is still considered valid (a
lled nish time would signify that the speech act is no longer
valid as it might be a retracted error).
Episode 2: John, John, absence, course 123, 1-6-90, 30-6-90, assert, assume,
,
Remarks: (a) Generally speaking, the agent who performs the speech act and
the one who does the action may be dierent. However, in this
particular example, John is the agent for both.
(b) John actually left on 1-6-90 (indicated by the start mood assert),
but intended to come back on 30-6-90 (by the nish mood
assume).
(c) The time of the performance of the speech act is not recorded,
and it is still a valid act (seen from the two unlled times).
Episode 3: Susan, John, absence, course 123, 1-6-90, 30-6-90, assume,
assume, now,
Remarks: (a) Susan makes a hypothesis now about Johns absence in the future
because it can be seen that both the start mood and the nish
mood are assume.
(b) The time of now must be earlier than 1-6-90 because the assump-
tion can only be made about the future, though there is no other
sign to prove the exact time in the speech.
Episode 4: Susan, John, absence, course 123, 1-6-90, , assume, , 1-6-90,
John, John, absence, course 123, 1-6-90, 30-6-90, plan, plan, 15-
5-90,
John, John, absence, course 123, 2-6-90, 30-6-90, assert, plan, 5-
6-90,
, John, absence, 2-6-90, 20-6-90, assert, assert, 21-6-90,
Remarks: (a) Susan made an assumption about Johns leaving date, and she did
not make any guess about his return date.
(b) John planned his leave from 1-6-90 till 30-6-90. Though he might
not be able to follow it, it should still be kept as a valid record in
the memory.
(c) The third expression says that though John did not leave as he
planned, he still planned to be back on 30-6-90.
(d) The last ST tells the actual period of Johns leave.
(e) The start times and nish times of performances of speech acts in
the four ST expressions are signicant; they record when the
speeches were uttered and whether or not they are still valid.
Episode 5: Peter, Susan, John, absence, course 123, 1-6-90, 30-6-90, ques-
tion, question, yesterday, , assert, , today,
Remarks: (a) Susan enquired as to both the start time and the nish time of
Johns leave, which was done yesterday.
(b) Peter reported the start time of Susans speech act, which was
done today.
(c) If an ST is like:
Peter, Susan, John, absence, 1-6-90, 30-6-90, question,
question, yesterday, , assert, , this morning, this
afternoon
then it means this morning it was recorded that Peter reported
198 Appendix A
Susans enquiry, but this afternoon it was found out that it was a
mistake Peter did not say anything about Susans question. In
short, the lled nish time afternoon would set the ST record
invalid.
A.3 Examples of surrogates
The surrogate structure is the basic record structure of the semantic temporal data-
base. The examples here show how the specications in ST can be translated into
surrogates in the semantic temporal database. The surrogate is dened as a tuple as
follows (it is rstly dened in section 10.2 in which one example of surrogate repre-
sentation with detailed explanations can be found):
surrogate :surrogate-identication, content, antecedents, authority,
authority, mood, mood, action-time, action-time, record-
time, record-time
All ve examples given below are translations of the corresponding episodes in ST
form in the last section into the surrogate representation (that is, Examples 1, 2, 3
and so on below correspond to Episodes 1, 2, 3 and so on of the last section).
Example 1:
Surrogate: srg-id1, [absence], John, course 123, , , assert, assert, 1-6-90, 30-
6-90, [today],
Remarks: (a) The surrogate identication, srg-id1, is supposed to be uniquely
generated by the technical database system.
(b) the item [absence] is a dened action type (an aordance dened
in the ontology chart) in the database.
(c) Antecedents John and course 123 relate the action being
absent to the context for determination of semantics.
(d) The two dashes represent the unknown authority for start and
authority for nish who reported or recognised the action Johns
absent from the course 123.
(e) Both the mood for start and mood for nish are assertive (assert)
which means the semantic content in this surrogate is a reported
observation.
(f) Two times 1-6-90 and 30-6-90 are the start action time and nish
action time to indicate the period of Johns absence.
(g) The item [today] is the start record time, which means the record-
ing of the speech act is today; [today] will be converted into the
date of the day when the surrogate is entered into the semantic
temporal database.
(h) The unlled nish record time signies the surrogate is still valid. If
the nish record time is lled, then it means the surrogate is
marked as an obsolete record.
Example 2: srg-id2, [absence], John, course 123, , , assert, plan, 1-6-90, 30-6-
90, [today],
Remarks: The values lled in the surrogate that the start action time is assert
and the nish action time is plan indicate that the start of Johns leave
on 1-6-90 is an observed fact and the nish of the leave is a projected
plan.
Example 3: srg-id3, [absence], John, course 123, Susan, Susan, assume, assume,
1-6-90, 30-6-90, [now],
Semantic templates and surrogate specication 199
Remarks: Susan is the responsible person who assumed the start and nish times
of Johns absence, therefore Susan has been put in the elds of start
authority and nish authority.
Example 4: srg-id41, [absence], John, course 123, Susan, , assume, , 1-6-90,
, 1-6-90,
srg-id42, [absence], John, course 123, John, John, plan, plan, 1-6-90,
30-6-90, 15-5-90,
srg-id43, [absence], John, course 123, , John, assert, plan, 2-6-90,
30-6-90, 5-6-90,
srg-id44, [absence], John, course 123, , , assert, assert, 2-6-90,
20-6-90, 21-6-90,
Remarks: Particular person names lled as authorities indicate who made the
observations or assumptions (to read this example in connection with
Narrative 4 and Episode 4 will enable the reader to understand the
example).
Example 5: srg-id5, [absence], John, course 123, Susan, Susan, question, ques-
tion, 1-6-90, 30-6-90, [today],
Remarks: Susan questioned whether or not John was o from 1-6-90 until 30-6-
90, which can be seen from the mood for start and mood for nish, both
being lled with question.
One of the purposes of discussing these examples is to illustrate a complete process,
which ranges from semantic analysis and modelling to semantic temporal database
design of a technical information system.
200 Appendix A
Appendix B LEGOL applications in the
CRIS case
The CRIS case was invented by IFIP (the International Federation for Information
Processing) for the comparative review of information systems design methodolo-
gies. The case denition can be found in Information Systems Design Methodologies
(Olle et al. 1982).
In this appendix, the CRIS case is studied using the method of Semantic Analysis
which results in an ontology chart (Figure B.1). In order to test LEGOL language
and the interpreter, questions are made and then translated into LEGOL expres-
sions. The LEGOL statements were executed by the LEGOL interpreter, a part of
an early prototype of the Normbase run on a VAX station, and the results were pro-
duced based on some imaginary data stored in the Normbase. Section B.1 contains
some questions and the LEGOL statements, and section B.2 presents the corre-
sponding output from the Normbase version.
B.1 Questions and LEGOL statements
In this section, there are questions concerning the CRIS case, and statements in
LEGOL that will produce the answers, printed below each question.
(1) What are the Technical Committees of the various learned societies on our
records?
TC
(2) Who are the current member organizations of IFIP ?
current membership(national-org, IFIP)
(3) What Working Groups now exist?
WG
or
WG(TC)
(4) When was WG8.1 started?
start-of WG# WG8.1
(5) Which are the newly created Working Groups after 1980?
start-of WGafter @1980
or
start-of WGafter start-of WG# WG8.1
(Note: the at sign (i.e. @) is an ad hoc solution for making that version of the
compiler distinguish what after the at sign will be a number for year.)
(6) Who were eligible to attend the Working Conference CRIS-3?
eligible(person, WC#CRIS-3)
201
m
e
m
b
e
r
#
p
r
i
o
r
i
t
y
c
i
t
i
z
e
n
m
e
m
b
e
r
n
a
t
.

r
e
p
m
e
m
b
e
r
a
t
t
e
n
d
a
n
t
c
o
n
t
r
i
b
u
t
o
r
#
s
e
l
e
c
t
e
d
m
e
m
b
e
r
o
r
g
a
n
i
s
a
t
i
o
n
#
I
F
I
P
n
a
t
i
o
n
a
l
-
o
r
g
m
e
m
b
e
r
s
h
i
p
m
e
m
b
e
r
s
h
i
p
c
i
t
i
z
e
n
s
h
i
p
n
a
t
i
o
n
s
o
c
i
e
t
y
W
C
T
C
W
G
a
r
e
a
e
l
i
g
i
b
l
e
m
e
e
t
i
n
g
s
u
b
j
e
c
t
r
e
p
r
e
s
e
n
t
a
t
i
o
n
m
e
m
b
e
r
s
h
i
p
m
e
m
b
e
r
s
h
i
p
m
e
m
b
e
r
s
h
i
p
s
p
o
n
s
o
r
s
o
n
o
n
r
e
v
i
e
w
c
o
n
t
r
i
b
u
t
i
o
n
a
t
t
e
n
d
a
n
c
e
w
o
r
k
a
b
s
t
r
a
c
t
p
a
p
e
r
r
e
p
o
r
t
c
o
m
m
i
t
t
e
e
P
C
O
C
t
o
p
e
r
s
o
n
m
e
m
b
e
r
F
i
g
u
r
e

B
.
1
.
O
n
t
o
l
o
g
y

c
h
a
r
t

o
f
t
h
e

C
R
I
S

c
a
s
e
.
W
C
w
o
r
k
i
n
g

c
o
n
f
e
r
e
n
c
e
;
T
C
t
e
c
h
n
i
c
a
l

c
o
m
m
i
t
t
e
e
;
W
G
w
o
r
k
i
n
g

g
r
o
u
p
;
P
C
p
r
o
g
r
a
m
m
e

c
o
m
m
i
t
t
e
e
;
O
C
o
r
g
a
n
i
s
i
n
g

c
o
m
m
i
t
t
e
e
.
(7) Who has ever contributed a paper in any area of interest to an IFIP confer-
ence?
contribution(person, paper) where on(paper, area)
(8) Who has ever contributed a paper on the subject of CRIS?
contribution(person, paper) where on(paper, subject#CRIS)
(9) Who has actually contributed a selected paper to a conference?
to(contribution(person, paper), WC#IFIP) where selected(paper)
(10) Who have the highest priority to attend the conference CRIS-3?
high priority(eligible(person, WC#CRIS-3))
(11) Who has ever attended an IFIP conference?
attendance(person, meeting)
(12) Who attended the working conference CRIS-3?
attendance(person, meeting#CRIS-3)
(13) Is there anyone who attended, but did not have the rst priority for the
Working Conference CRIS-3?
attendance(person, meeting#CRIS-3) while highest priority(eligible(person,
WC#CRIS#3))
(14) Who are the sponsors of the Working Conference CRIS-3?
sponsorship(WG, WC#CRIS-3)
(15) Who are members of sponsoring working groups but did not attend the con-
ference?
membership(person, WG) while sponsorship(WG, WC#CRIS-3) while-not
B.2 Output from the Normbase
This section shows the output of the Normbase prototype which runs on a VAX
station. The results listed below answer the questions of section B.1 with the corre-
sponding numbers.
1.
LegolTC
TC start nish
TC8 1974-01-01 -
TC7 1972-01-01 -
TC2 1965-01-01 -
2.
Legolcurrent membership(national-org, IFIP)
membership national-org IFIP start nish
NGI IFIP 1962-01-01 -
RSA IFIP 1979-01-01 -
DVI IFIP 1990-10-06 -
Nsk Dataforen. IFIP 1975-01-01 -
BCS IFIP 1960-01-01 -
A.F. Cyb Econ. IFIP 1960-01-01 -
and
Legolpast membership(national-org, IFIP)
membership national-org IFIP start nish11 11
DVI IFIP 1960-01-01 1990-10-05
LEGOL applications in the CRIS case 203
3.
LegolWG
WG start nish111 1
WG2.1 (Algol) 1960-06-01 -
WG8.4 1958-01-01 -
WG8.3 (DSS) 1981-01-01 -
WG2.4 1973-01-01 -
WG2.3 1969-07-01 -
WG2.2 1965-05-01 -
WG2.5 1974-07-01 -
WG8.1 1976-01-01 -
WG7.1 1972-01-01 -
WG2.6 1974-01-01 1985-12-31
or
LegolWG(TC)
WG TC start nish111 1
WG2.1 (Algol) TC2 1960-06-01 -
WG8.4 TC8 1985-01-01 -
WG8.3 (DSS) TC8 1981-01-01 -
WG2.4 TC2 1973-01-01 -
WG2.3 TC2 1969-07-01 -
WG2.2 TC2 1965-05-01 -
WG2.5 TC2 1974-07-01 -
WG8.1 TC8 1976-01-01 -
WG7.1 TC7 1972-01-01 -
WG2.6 TC2 1974-01-01 1985-12-31
4.
Legolstart-of WG#WG8.1
WG start nish
WG8.1 1976-01-01 1976-01-01
otherwise
LegolWG#WG8.1
WG start nish
WG8.1 1976-01-01 -
5.
Legolstart-of WG after @1980
WG start nish111 1
WG8.4 1985-01-01 1985-01-01
WG8.3 (DSS) 1981-01-01 1981-01-01
204 Appendix B
6.
Legoleligible(person, WC#CRIS-3)
eligible person WC start nish
L. Penedo CRIS-3 1988-05-01 -
R. Piloty CRIS-3 1988-05-01 -
A. Finkelstein CRIS-3 1988-05-01 -
J. Hagelstein CRIS-3 1988-05-01 -
J. Fourot CRIS-3 1988-05-01 -
Alex V. Stuart CRIS-3 1988-05-01 -
A. Bertztiss CRIS-3 1988-05-01 -
Bill Ollie CRIS-3 1988-05-01 -
R.K. Stamper CRIS-3 1988-05-01 -
B.L. Sendor CRIS-3 1988-05-01 -
7.
Legolcontribution(person, paper) where on(paper, area)
contribution person paper on area start nish
L. Penedo EDI & Groupware Groupware 1990-06-01 -
J. Hagelstein Groupware Groupware 1990-06-01 -
J. Fourot Prog. in Algol Struct. Prog. 1985-01-01 -
Alex V. Stuart Temporal DBMS Temp. databases 1980-07-01 -
A. Bertztiss Time in Data Temp. databases 1986-06-01 -
Bill Ollie New IS Analysis CRIS 1988-06-01 -
R.K. Stamper LEGOL 2.1 Temp. databases 1976-01-01 -
R.K. Stamper DB vs Normbase Groupware 1990-05-10 -
B.L. Sendor CRIS Review CRIS 1989-12-15 -
8.
Legolcontribution(person, paper) where on(paper, subject#CRIS)
contribution person paper on subject start nish
Bill Ollie New IS Analysis CRIS 1988-06-01 -
R.K. Stamper LEGOL 2.1 CRIS 1976-01-01 -
R.K. Stamper MEASUR CRIS 1976-01-01 -
R.K. Stamper DB vs Normbase CRIS 1990-05-10 -
B.L. Sendor CRIS Review CRIS 1989-12-15 -
9.
Legolto( contribution(person, paper), WC(IFIP) ) where selected(paper)
contri-
bution person paper WC IFIP selected start nish
- L. Penedo EDI & Groupware COCIS-91 IFIP - 1990-06-01 -
- J. Hagelstein Groupware COCIS-91 IFIP - 1990-06-01 -
- J. Fourot Prog. in Algol Structured Prog. IFIP - 1985-01-01 -
- Alex V. Stuart Temporal DBMS Temp. asp. IS IFIP - 1980-07-01 -
- A. Bertztiss Time in Data Temp. asp. IS IFIP - 1986-06-01 -
- Bill Ollie New IS Analysis CRIS-3 IFIP - 1988-06-01 -
- R.K. Stamper LEGOL 2.1 Temp. asp. IS IFIP - 1976-01-01 -
- R.K. Stamper DB vs Normbase COCIS-91 IFIP - 1990-05-10 -
- B.L. Sendor CRIS Review CRIS-3 IFIP - 1989-12-15 -
10.
Legolpriority(eligible(person, WC#CRIS#3))
priority eligible person WC start nish
2 R. Piloty CRIS-3 1988-05-01 -
1 Alex V. Stuart CRIS-3 1988-05-01 -
1 J. Fourot CRIS-3 1988-06-01 -
1 R.K. Stamper CRIS-3 1988-06-01 -
1 Bill Ollie CRIS-3 1988-05-01 -
and
Legollowest priority(eligible(person, WC#CRIS-3))
priority eligible person WC start nish
1 Alex V. Stuart CRIS-3 1988-05-01 -
1 J. Fourot CRIS-3 1988-06-01 -
1 R.K. Stamper CRIS-3 1988-06-01 -
1 Bill Ollie CRIS-3 1988-05-01 -
11.
Legolattendance(person, meeting)
attendance person meeting start nish111 1
F. Bodart CRIS-3 1990-11-26 1990-11-30
A. Finkelstein CRIS-3 1990-11-26 1990-11-30
B.L. Sendor CRIS-3 1990-11-26 1990-11-30
R.K. Stamper COSCIS-91 1991-08-27 1991-08-29
G. Bracchi CRIS-3 1990-11-28 1990-11-30
E. Falkenberg CRIS-3 1990-11-26 1990-11-27
R.K. Stamper CRIS-3 1990-11-26 1990-11-30
H.W. Le Roux CRIS-3 1990-11-26 1990-11-30
Bill Ollie CRIS-3 1990-11-26 1990-11-30
R. Piloty CRIS-3 1990-11-27 1990-11-30
J. Fourot CRIS-3 1990-11-26 1990-11-30
A.A. Dorodnicyn Parallel Proc. 1986-03-04 1986-03-05
A. Bertztiss Temp. asp. IS 1988-09-12 1988-09-16
A. Finkelstein Temp. asp. IS 1988-09-12 1988-09-15
O. Longe Parallel Proc. 1986-03-04 1986-03-03
L. Penedo Parallel Proc. 1986-03-04 1986-03-03
F. Bodart Parallel Proc. 1986-03-04 1986-03-05
H.W. Le Roux Temp. asp. IS 1988-09-13 1988-09-16
B. Gunadi Parallel Proc. 1986-03-04 1986-03-05
H.W. Le Roux Parallel Proc. 1986-03-04 1986-03-05
206 Appendix B
12.
Legolattendance(person, meeting#CRIS-3)
attendance person meeting start nish111 1
F. Bodart CRIS-3 1990-11-26 1990-11-30
A. Finkelstein CRIS-3 1990-11-26 1990-11-30
B.L. Sendor CRIS-3 1990-11-26 1990-11-30
G. Bracchi CRIS-3 1990-11-28 1990-11-30
E. Falkenberg CRIS-3 1990-11-26 1990-11-27
R.K. Stamper CRIS-3 1990-11-26 1990-11-30
H.W. Le Roux CRIS-3 1990-11-26 1990-11-30
Bill Ollie CRIS-3 1990-11-26 1990-11-30
R. Piloty CRIS-3 1990-11-27 1990-11-30
J. Fourot CRIS-3 1990-11-26 1990-11-30
13.
Legolattendance(person, meeting#CRIS-3) while highest
priority(eligible(person, WC#CRIS-3))
attendance person meeting priority eligible WC start nish111 1
R. Piloty CRIS-3 2 CRIS-3 1990-11-27 1990-11-30
14.
Legolsponsorship(WG, WC)
sponsorship WG WC start nish
WG8.1 Temp. asp. IS 1986-06-01 1989-01-01
WG2.1 (Algol) Structured Prog. 1984-01-01 1988-01-01
WG8.1 COSCIS-91 1990-01-01 1992-12-31
WG8.1 CRIS-3 1988-01-01 -
and
Legolsponsorship(WG, WC#CRIS-3)
sponsorship WG WC start nish
WG8.1 CRIS-3 1988-01-01 -
15
Legolmembership(person, WG) while sponsorship(WG, WC#CRIS-3) while-not
member- spon- atten-
ship person WG sorship WC dance meeting start nish111 1
F. Bodart WG8.1 CRIS-3 CRIS-3 1990-11-26 1990-11-26
F. Bodart WG8.1 CRIS-3 CRIS-3 1990-11-29 1990-11-30
Bill Ollie WG8.1 CRIS-3 CRIS-3 1990-11-26 1990-11-26
Bill Ollie WG8.1 CRIS-3 CRIS-3 1990-11-29 1990-11-30
R.K. Stamper WG8.1 CRIS-3 CRIS-3 1990-11-28 1990-11-30
Bibliography
Achueler, B.-M., 1977, Update reconsidered. In Nijssen, G. M. (ed.), Architecture
and Models in Data Base Management Systems. North-Holland, Amsterdam.
Acko, R., 1967, Management misinformation systems. Management Science, 14
(4), B14756.
Adema, J. J., 1990, Models and Algorithms for the Construction of Achievement
Tests. Doctoral Thesis, University of Twente, Enschede.
Ades, Y. M., 1989, NAMAT System User Manual. SIS project, University of
Qatar.
Aho, A. V., Sethi, R. and Ullman, J. D., 1986, Compilers: Principles, Techniques,
and Tools. Addison-Wesley, Reading, Massachusetts.
Allen, L. E. and Saxon, C. S., 1986, Analysis of the logical structure of legal rules
by a modernized and formalized version of Hohfelds fundamental legal
conceptions. In Martino, A. A. and Natali, F. S. (eds.), Automated Analysis of
Legal Texts: Logic, Informatics, Law. Elsevier Science, Amsterdam, 385451.
Andersen, P. B., 1990, A Theory of Computer Semiotics: Semiotic Approaches to
Construction and Assessment of Computer Systems. Cambridge University
Press, Cambridge.
Andersen, P. B., 1991, A semiotic approach to construction and assessment of
computer systems. In Nissen, H. E., Klein, H. K. and Hirschheim, R. (eds.),
Information Systems Research: Contemporary Approaches and Emergent
Traditions. Elsevier Science, North-Holland, 465514.
qvist, L., 1984, Deontic logic. In Gabbay, D. and Guenthner, D. (eds.),
Handbook of Philosophical Logic, Volume II. D. Reidel Publishing Company,
Dordrecht, Netherlands, 605714.
Atkinson, M., Bancilhon, F., DeWitt, D., Dittrich, K., Maier, D. and Zdonik, S.,
1990, The object-oriented database system manifesto. Proceedings of DPPD
Conference, Kyoto, Japan, December 1989, 4057.
Auramki, E., Lehtinen, E. and Lyytinen, K., 1988, A speech-act-based oce
modeling approach. ACM Transactions on Oce Information Systems, 6 (2),
12652.
Austin, J. L., 1980, How to Do Things with Words, the William James Lectures
delivered at Harvard University in 1955. Oxford University Press, Oxford.
Avison, D. E. and Fitzgerald, G., 1995, Information Systems Development:
Methodologies, Techniques and Tools. McGraw-Hill, London.
Avison, D. E. and Nandhakumar, J., 1995, The discipline of information systems:
208
let many owers bloom. In Falkenberg, E. D., Hesse, W. and Olive, A. (eds.)
Information Systems Concepts: towards a Consolidation of Views
(Proceedings of the IFIPWG8.1 International Conferences), Chapman and
Hall, London, pp. 117.
Barbic, F. and Pernici, B., 1985, Time Modelling in Oce Information Systems.
In Proceedings of ACM-SIGMOD, International Conference on Management
of Data.
Beck, K. and Cunningham, W., 1989, A laboratory for teaching object-oriented
thinking. In Megrowitz, N. K. (ed.), OOPSLA89 Proceedings, New Orleans,
Oct. 16.
Benyon, D., 1994, A functional model of interacting systems: a semiotic
approach. In Connolly, J. H. and Edmonds, E. A. (eds.), CSCW and Articial
Intelligence, Springer-Verlag, London.
Bickerton, M. and Siddiqi, J., 1993, The classication of requirements
engineering methods. Proceedings RE93, IEEE International Symposium on
Requirements Engineering. San Diego.
Boehm, B., 1981, Software Engineering Economics. Prentice-Hall, Englewood
Clis, New Jersey.
Boekkooi-Timminga, E., 1989, Models for Computerised Test Construction.
Doctoral dissertation, University of Twente. Academisch Boeken Centrum,
De Lier, Netherlands.
Boekkooi-Timminga, E. and Sun, L., 1991, Contest: a computerised test
construction system. In Hoogstraten, J. and Linden, W.J. v., (eds.),
Proceedings of Dutch National Conference on Educational Research
Onderwijs-researchdagen91. Stichting Centrum voor Onderwijsonderzoek
van de Universiteit van Amsterdam, Amsterdam, 6976.
Booch, G., 1994, Object Oriented Design, with Applications. The
Benjamin/Cummings Publishing Company, Redwood City, California.
Booch, G., Rumbaugh, J. and Jacobson, I., 1998, Unied Modelling Language
User Guide. Addison-Wesley, Reading, Massachusetts.
Boustred, C., 1997, ATSI Seminars, http://www.strategize.com/presentations/
paradigm/sld014.htm (accessed on 28 February 1998).
Brachman, R. J., 1979, On the epistemological status of semantic networks. In
Findler, N. V. (ed.), Associative Networks: Representation and Use of
Knowledge by Computers. Academic Press, New York, 350.
Brachman, R. J. and Levesque, H. J. (eds.), 1985, Readings in Knowledge
Representation. Morgan Kaufmann, Los Altos, California.
Bradley, R. and Swartz, N., 1979, Possible Worlds: an Introduction to Logic and Its
Philosophy. Basil Blackwell Publisher, Oxford.
Brkic, J., 1970, Norm and Order an Investigation into Logic, Semantics, and the
Theory of Law and Morals. Humanities Press, New York.
Burrell, G. and Morgan, G., 1979, Sociological Paradigms and Organisational
Analysis. Heinemann, London.
Checkland, P. B., 1981, Systems Thinking, Systems Practice. Wiley, Chichester.
Chellas, B. F., 1980, Modal Logic: an Introduction. Cambridge University Press,
Cambridge.
Chen, P. P., 1976, The entityrelationship model, towards a unied view of data.
ACM Transactions on Database Systems, 1 (1), 936.
Cliord, J. and Tansel A. U., 1985, On an algebra for historical relational
databases: two views. In Proceedings of ACM-SIGMOD, International
Conference on Management of Data.
Bibliography 209
Coad, P. and Yourdon, E., 1990, Object-Oriented Analysis. Prentice-Hall,
Englewood Clis, New Jersey.
Codd, E. F., 1970, A relational model of data for large shared data banks.
Communications of the ACM, 13 (6), 37787.
Cunningham, Donald J., 1992, Beyond educational psychology: steps toward an
educational semiotics. Educational Psychology Review, 4, 16594.
Date, C. J., 1995, An Introduction to Database Systems, Volume I (Fifth Edition).
Addison-Wesley, Reading, Massachusetts.
Devlin, K., 1991, Logic and Information, Cambridge University Press,
Cambridge.
Dietz, J. L. G., 1992, Subject-oriented modelling of open active systems. In
Falkenberg, E.D., Rolland, C. and El-Sayed, E. N. (eds.), Information System
Concepts: Improving the Understanding. Elsevier Science, Amsterdam,
227338.
Dietz, J. L. G. and Widdershoven, G. A. M., 1991, Speech acts or communicative
action? In Bannon, L., Robinson, M. and Schmidt, K. (eds.), Proceedings of
the Second European Conference on Computer-Supported Cooperative Work
(CSCW91). Kluwer Academic Publishers, Dordrecht, Netherlands,
23548.
Dijk, T. A. v., 1981, Studies in the Pragmatics of Discourse. Mouton Publishers,
The Hague.
Dik, S. C., 1979, Functional Grammar. North-Holland, Amsterdam.
Dik, S. C., 1989, The Theory of Functional Grammar, Part 1: the Structure of the
Clause. Foris Publications, Dordrecht, Netherlands.
Eemeren, F. H. v. and Grootendorst, R., 1984, Speech Acts in Argumentative
Discussions. Foris Publications, Dordrecht, Netherlands.
Embley, D. W., Kurtz, B. D. and Woodeld, S. N., 1992, Object-Oriented Systems
Analysis, a Model-Driven Approach. Prentice Hall, Englewood Clis, New
Jersey.
Falkenberg, E.D., Hesse, W., Lindgreen, P., Nilsson, B.E., Oei, J.L.H., Rolland,
C., Stamper, R.K., Assche, F.J.M.V, Verijn-Stuart, A.A., Voss, K. 1998, A
framework of information system concepts, The FRISCO Report (Web
edition), ftp://p.leidenuniv.nl/pub/rul/fri-full.zip (accessed 10/12/1998).
Fiske, J., 1990, Introduction to Communication Studies. Routledge, London.
Flores, F., Graves, M., Harteld, B. and Winograd, T., 1988, Computer systems
and the design of organisational interaction. ACM Transactions on Oce
Information Systems, 6 (2), 15372.
Follesdal, D. and Hilpinen, R., 1970, Deontic logic: an introduction. In Hilpinen,
R. (ed.), Deontic Logic: Introductory and Systematic Readings. D. Reidel
Publishing Company, Dordrecht, Netherlands, 135.
Franke. R., 1987, Technological revolution and productivity decline: computer
introduction in the nancial industry. Technological Forecasting and Social
Change, 31, 14354.
Gadia, S. K. and Vaishnav J. H., 1985, A query language for a homogeneous
temporal database. In Proceedings of the Fourth ACM SIGACT-SIGMOD
Symposium on Principles of Database Systems.
Galliers, R. (ed.), 1987, Information Analysis Selected Readings. Addison-
Wesley, Wokingham, Berkshire, and Sydney.
Gibbs, J. P., 1981, Norms, Deviance, and Social Control Conceptual Matters.
Elsevier, New York.
Gibson, J. J., 1968, The Ecological Approach to Visual Perception. Houghton
Miin Company, Boston, Massachusetts.
210 Bibliography
Goguen, J., 1992, The Dry and the Wet. In Falkenberg, E. D., Rolland, C. and
El-Sayed, E. N. (eds.), Information System Concepts: Improving the
Understanding. Elsevier Science, Amsterdam, 118.
Gonzalez, R., 1997, Hypermedia data modeling, coding, and semiotics.
Proceedings of the FEEE, 85(7), 111140.
Griethuysen, J. J. v. (ed.), 1982, Concepts and Terminology for the Conceptual
Schema and the Information Base, ISO/TC97/SC5-N695.
Groot, C. de, 1992, Pragmatics in functional grammar. Proceedings of Workshop
of TWLT 4. University of Twente, Enschede.
Haack, S., 1978, Philosophy of Logics. Cambridge University Press, Cambridge.
Hawkes, T., 1977, Structuralism and Semiotics. Routledge, London.
Hilpinen, R. (ed.), 1971, Deontic Logic: Introductory and Systematic Readings. D.
Reidel Publishing Company, Dordrecht, Netherlands.
Hirschheim, R., 1985, Oce Automation: a Social and Organizational Perspective.
John Wiley and Sons, Chichester.
Hirschheim, R. and Klein, H. K., 1989, Four paradigms of information systems
development. Communications of the ACM, 32 (10), 11991216.
Hirschheim, R., Klein, H. K. and Lyytinen, K., 1995, Information Systems
Development and Data Modelling: Conceptual and Philosophical Foundations.
Cambridge University Press, Cambridge.
Hoede, C., 1986, Similarity in Knowledge Graphs. Department of Applied
Mathematics, University of Twente, Enschede.
Holy, L. and Stuchlik, M., 1983, Actions, Norms and Representation, Foundations
of Anthropological Inquiry. Cambridge University Press, Cambridge.
Howe, D. R., 1989, Data Analysis for Database Design. Routledge, Chapman and
Hall, London.
Huang, K., 1998, Organisational Aspect of EDI: a Norm-Oriented Approach. PhD
thesis, University of Twente, Enschede.
Jackman, M. and Pavelin, C., 1988, Conceptual graphs. In Rinland, G. A. and
Duce, D. A. (eds.), Approaches to Knowledge Representation. Research
Studies Press Ltd, Taunton, Somerset, and John Wiley and Sons, New York,
16174.
Jackson, M. A., 1983, Systems Development. Prentice Hall, Englewood Clis,
New Jersey.
Jayaratna, N., 1986, Normative Information Model-based Systems Analysis and
Design (NIMSAD): a framework for understanding and evaluating
methodologies. Journal of Applied Systems Analysis, 13, 7387.
Jayaratna, N., 1990, Systems analysis: the need for a better understanding.
International Journal of Information Management, 10, 22834.
Jones, A. J. I., 1983, Communication and Meaning, an Essay in Applied Modal
Logic. D. Reidel Publishing Company, Dordrecht, Netherlands.
Jones, S., Mason, P. and Stamper, R., 1979, LEGOL 2.0: a relational specication
language for complex rules. Information Systems, 4, 2848.
Kim, J., Yoo, H. and Lee, Y., 1990, Design and implementation of a temporal
query language with abstract time. Information Systems, 15 (3), 34957.
Klarenberg, P., 1991, Issues on Design and Implementation of the Normbase
Systems. Thesis, University of Twente, Enschede.
Klein, H. K., 1996, Preface: the potential contribution of semiotics and systems
theory to the continuing evolution of information systems research. In
Holmqvist, B., Andersen, P. B., Klein, H. and Posner, R. (eds.), Signs of
Work: Semiosis and Information Processing in Organisations. Walter de
Gruyter, Berlin, vviii.
Bibliography 211
Kolkman, M., 1993, Problem Articulation Methodology. Doctoral Thesis,
University of Twente, Enschede.
Kolkman, M., 1995, Managing ambiguity and change with the problem
articulation methodology. In Pettigrew A. and Hines D. (eds.), Globalisation
and Process Analysis of the Organisation. Sage, 00000.
Kolkman, M. and Liu, K., 1991, Systems physiology: an approach of norm-
governed action. In Harten, A. V. and Pol, B. G. F. (eds.), Proceedings of
Dutch National Conference of Business Research (NOBO). Enschede.
Febodruk, 27584.
Kotonya, G. and Sommerville I., 1998, Requirements Engineering: Processes and
Techniques. John Wiley and Sons, New York.
Kuhn, T. S., 1962, The Structure of Scientic Revolutions (2nd edition). University
of Chicago Press, Chicago.
Lacey, A. R., 1976, A Dictionary of Philosophy. Routledge and Kegan Paul,
London.
Lakatos, I., 1970, Falsication and the methodology of scientic research
programmes. In Lakatos, I. and Musgrave, A. (eds.), Criticism and the
Growth of Knowledge. Cambridge University Press, Cambridge, 91196.
Lee, R. M., 1988, Bureaucracies as deontic systems. ACM Transactions on Oce
Leslie, R. E., 1986, Systems Analysis and Design. Prentice-Hall, Englewood Clis,
New Jersey.
Lindgreen, P. (ed.), 1990, A Framework of Information Systems Concepts (interim
report). FRISCO Task Group, IFIP WG 8.1.
Liu, K. 1993, Semiotics Applied to Information Systems Development. PhD thesis,
Liu, K., Ades, Y. and Stamper, R., 1994, Simplicity, uniformity and quality: the
role of Semantic Analysis in systems development. In Ross, M., Brebbia, C.
A., Staples, G. and Stapleton, J. (eds.), Software Quality Management.
Computational Mechanics Publications, 2, 21935.
Liu, K. and Dix, A., 1997, Norm governed agents in CSCW. The First
International Workshop on Computational Semiotics. Paris. University of De
Vince.
Liu, K. and Gao, Z. L., 1997, A problem articulation method for information
systems: planning for responsive information management infrastructure. In
Avison, D. (ed.), Key Issues in Information Systems. McGraw-Hill,
Maidenhead, Berkshire, 35362.
Loomis, M. E. S., 1990, OODBMS vs. relational. Journal of Object Oriented
Programming, July/August, 7982.
Loux, M. J., 1978, Substance and Attribute a Study in Ontology. D. Reidel
Publishing Company, Dordrecht, Netherlands.
Lundeberg, M., Goldkuhl, G. and Nilsson, A., 1981, Information Systems
Development: a Systematic Approach. Prentice-Hall, Englewood Clis, New
Jersey.
Lyons, J., 1977, Semantics, Volume I. Cambridge University Press, Cambridge.
Lyytinen, K., 1987, Two views of information modeling. Information and
Management, 12, 919.
Lyytinen, K. and Lehtinen, E., 1986, Action based model of information system.
Machlup, F., 1980, Knowledge: Its Creation, Distribution, and Economic
Signicance Volume I: Knowledge and Knowledge Production. Princeton
University Press, Princeton, New Jersey.
212 Bibliography
Macro, A. and Buxton, J., 1987, The Craft of Software Engineering. Addison-
Wesley, Wokingham, Berkshire.
Masterman, M., 1970, The nature of a paradigm. In Lakatos, I. and Musgrave,
A. (eds.),Criticism and the Growth of Knowledge. Cambridge University
Press, Cambridge, 5889.
McCarty, L. T., 1986, Permissions and obligations: an informal introduction. In
Martino, A. A. and Natali, F. S. (eds.), Automated Analysis of Legal Texts:
Logic, Informatics, Law. Elsevier Science, Amsterdam, 30737.
Meyer, J.-J. C., 1988, A dierent approach to deontic logic: deontic logic viewed
as a variant of dynamic logic. Notre Dame Journal of Formal Logic, 29 (1),
10936.
Meyer, J.-J. C. and R. J. Wieringa, 1993, Deontic logic: a concise overview. In
Meyer, J.-J. C. and Wieringa, R. J. (eds.), Deontic Logic in Computer Science.
John Wiley and Sons, Chichester, 316.
Micallef, J., 1988, Encapsulation, reusability and extensibility in object-oriented
programming languages. Journal of Object-Oriented Programming,
April/May, 1234.
Michaels, C. F. and Carello, C., 1981, Direct Perception. Prentice-Hall,
Minker, J. (ed.), 1988, Foundations of Deductive Databases and Logic
Programming. Morgan Kaufmann, Los Altos, California.
Mishra, G. N., 1982, Ontology. Shri Maheshwari, Varanasi.
Morris, C. W., 1938, Foundations of the theory of signs. International
Encyclopedia of Unied Science, 1 (2). University of Chicago Press, Chicago.
Morris, C. W., 1946, Signs, Language and Behaviour. Braziller, New York.
Mumford, E. and Weir, M., 1979, Computer Systems in Work Design the
ETHICS Method. Associated Business Press, London.
Nachmias, C. and Nachmias, D., 1981, Research Methods in the Social Sciences.
St Martins Press, London.
Nauta, D., 1972, The Meaning of Information. Mouton, The Hague.
Nijssen, G. M. and Halpin, T. A., 1989, Conceptual Schema and Relational
Database Design, Prentice-Hall, Sydney.
Olle, T. W., Sol, H. G., and Verrijn-Stuart, A. A. (eds.), 1982, Information Systems
Design Methodologies: a Comparative Review. Elsevier Science, Amsterdam.
Olle, T. W., Hagelstein, J., Macdonald, I. G., Rolland, C., Sol, H. G., Assche, F. J.
M. v. and Verrijn-Stuart, A. A., 1991, Information Systems Methodologies: a
Framework for Understanding. Addison-Wesley, Wokingham, Berkshire.
OSW, 1995, The circulation document. Organisational Semiotics Workshop.
Enschede.
Pavelin, C., 1988, Logic in knowledge representation. In Ringland, G. A. and
Duce, D. A. (eds.), Approaches to Knowledge Representation: an Introduction.
Research Studies Press, Taunton, Somerset, 1344.
Peirce, C. S., 193158, ed. Hartshorne, C. and Weiss, P., Collected Papers of C. S.
Peirce. Harvard University Press, Cambridge, Massachusetts.
Polovina, A. and Heaton, J., 1992, An introduction to conceptual graphs. AI
Expert, May.
Popper, K., 1970, Normal science and its dangers. In Lakatos, I. and Musgrave,
A. (eds.), Criticism and the Growth of Knowledge. Cambridge University
Press, Cambridge, 518.
Porat, M. U., 1977, The Information Economy: Denition and Measurement. US
Department of Commerce, Washington DC.
Randal, D. M., 1988, Semantic networks. In Ringland, G. A. and Duce, D. A.,
Bibliography 213
(eds.), Approaches to Knowledge Representation: an Introduction. Research
Studies Press, Taunton, Somerset, 4580.
Reiter, R., 1984, Towards a logical reconstruction of relational database theory.
In Brodie, M. L., Mylopoulos, J. and Schmidt, J. W., (eds.), On Conceptual
Modelling: Perspectives from Articial Intelligence, Databases, and
Programming Languages. Springer-Verlag, New York, 191233.
Ringland, G. A. and Duce, D. A. (eds.), 1988, Approaches to Knowledge
Representation: an Introduction. Research Studies Press, Taunton, Somerset.
Ryle, G., 1949, The Concept of Mind. Hutchinson of London, London.
Ryu, Y. U. and Lee, R. M., 1993, Defeasible deontic reasoning: a logic
programming model. In Meyer, J.-J. C. and Wieringa, R. J. (eds.), Deontic
Logic in Computer Science. John Wiley and Sons, Chichester, 22541.
Sadeghi, R., Samson, W. B. and Deen, S. M., 1988, HQL a historical query
language. In Proceedings of the South British National Conference on
Databases (BNCOD6). University College, Cardi, 6986.
Sarda, N. L., 1990, Algebra and query language for a historical data model. The
Computer Journal, 33 (1), 1118.
Schank, R. C., 1972, Conceptual dependency: a theory of natural language
understanding. Cognitive Psychology, 3 (4).
Schank, R. C., 1975, Conceptual Information Processing. North-Holland.
Amsterdam.
Searle, J. R., 1969, Speech Acts an Essay in the Philosophy of Language.
Searle, J. R., 1979, Expression and Meaning. Cambridge University Press,
Cambridge.
Searle, J. R., 1983, Intentionality an Essay in the Philosophy of Mind. Cambridge
University Press, Cambridge.
Searle, J. R. and Vanderveken, D., 1985, Foundations of Illocutionary Logic.
Shannon, C. E. and Weaver W., 1949, The Mathematical Theory of
Communication. University of Illinois Press, Urbana, Illinois.
Shearer, I., 1992, Retrospective update: data as it was believed to be. The
Computer Journal, 35 (2), 1846.
Shlaer, S. and Mellor, S. J., 1988, Object Oriented Analysis, Modelling the World in
Data. Yourdon Press, Englewood Clis, New Jersey.
Shlaer, S. and Mellor, S. J., 1992, Object Lifecycles, Modelling the World in States.
Yourdon Press, Englewood Clis, New Jersey.
Smit, H. J., 1991, Consistency and Robustness of Knowledge Graphs. Doctoral
thesis, University of Twente, Enschede.
Snodgrass, R., 1985, A taxonomy of time in databases. Proceedings of ACM-
SIGMOD, International Conference on Management of Data.
Snodgrass, R., 1987, The temporal query language TQUEL. TODS, 12 (2),
24798.
Sowa, J., 1984, Conceptual Structures: Information Processing in Mind and
Machine. Addison-Wesley, Reading, Massachusetts.
Stamper, R. K., 1973, Information in Business and Administrative Systems. John
Wiley and Sons, New York.
Stamper, R. K., 1980, LEGOL: modelling legal rules by computer. In Niblett, B.
(ed.), Computer Science and Law; an Advanced Course. Cambridge
University Press, Cambridge, 127.
Stamper, R. K., 1985, Knowledge as action: a logic of social norms and
individual aordances. In Gilbert, G. N. and Heath, C., (eds.), Social Action
and Articial Intelligence. Gower Press, Aldershot, Hampshire, 17291.
214 Bibliography
Stamper, R. K., 1992, Language and computer in organised behaviour. In Riet,
R. P. v. d. and Meersman, R. A., (eds.), Linguistic Instruments in Knowledge
Engineering. Elsevier Science, Amsterdam, 14363.
Stamper, R. K., 1993, Social norms in requirements analysis an outline of
MEASUR. In Jirotka, M., Goguen, J. and Bickerton, M. (eds.),
Requirements Engineering, Technical and Social Aspects. Academic Press,
New York.
Stamper, R. K., Althaus, K., Backhouse, J., 1988, MEASUR: Method for
Eliciting, Analyzing and Specifying User Requirements. In Olle, T. W.,
Verrijn-Stuart, A. A. and Bhabuts, L., (eds.), Computerized Assistance
During the Information Systems Life Cycle. Elsevier Science, Amsterdam,
67116.
Stamper, R. K., Liu, K. and Huang, K., 1994, Organisational Morphology in Re-
engineering. The Second European Conference on Information Systems,
Nijenrode, 72937.
Stamper, R. K., Liu, K., Kolkman, M., Klarenberg, P., Slooten, V. F., Ades, Y.,
and Slooten, V. C., 1991, From database to normbase. International Journal
of Information Management, 11, 6279.
Stamper, R. K., Liu, K., Hafkamp, M. and Ades, Y., 1997, Signs plus norms
one paradigm for organisational semiotics. The First International
Workshop on Computational Semiotics, Paris.
Stamper, R. K. and Nauta, D., 1990, Information in a Semiotic Framework.
FRISCO Task Group of IFIP WG 8.1.
Stonebraker, M., Rowe, L. A., Lindsay, B., Gray, J., Carey, M., Brodie, M.,
Bernstein, Ph. and Beech, D., 1990, Third Generation Data Base System
Manifesto. The Committee for Advanced DBMS Function, Memorandum
No. UCB/ERL M90/29, College of Engineering, University of California,
Berkeley, California.
Stowell, F. (ed.), 1995, Information Systems Provision. McGraw-Hill, London.
Strassmann, P., 1980, The oce of the future: information management for the
new age. Technology Review, December/January.
Strassmann, P., 1990, The Business Value of Computers: an Executives Guide. The
Information Economics Press, New Canaan, Connecticut.
Thayse, A. (ed.), 1989, From Modal Logic to Deductive Databases. John Wiley
and Sons, Chichester.
Thnissen, K., 1990, Semantic Analysis: a Study and Comparison. MSc Thesis,
UKAIS, 1996, UK Academy for Information Systems Newsletter, 2 (1).
Verheijen, G. and Bekkum, J., 1982, NIAM: an information analysis method.
In Olle, W., Sol, H. and Verrijn-Stuart, A. (eds.), Information Systems
Design Methodologies: a Comparative Review. Elsevier Science, Amsterdam,
53790.
Way, E. C., 1991, Knowledge Representation and Metaphor. Kluwer Academic
Publishers, Dordrecht, Netherlands.
Weg, R. v. d. and Engmann, R., 1992, A framework and method for object-
oriented information systems analysis and design. In Falkenberg, E. D.,
Rolland, C. and El-Sayer, E. N. (eds.), Information System Concepts:
Improving the Understanding. Elsevier Science, Amsterdam, 12346.
Weigand, H., 1990, Linguistically Motivated Principles of Knowledge Base
Systems. Foris Publications, Dordrecht, Netherlands.
Weigand, H., 1993, Deontic aspects of communication. In Meyer, J.-J. Ch. and
Wieringa, R. J. (eds.), Deontic Logic In Computer Science. John Wiley and
Sons, Chichester, 25973.
Bibliography 215
Wieringa, R., Meyer, J.-J. C., and Weigand, H., 1989, Specifying dynamic and
deontic integrity constraints. Data and Knowledge Engineering, 4, 15790.
Wieringa, R. J. and Meyer, J.-J. C., 1993, Applications of deontic logic to
computer science: a concise overview. In Meyer, J.-J. Ch. and Wieringa, R. J.
(eds.), Deontic Logic in Computer Science. John Wiley and Sons, Chichester,
1740.
Wilks, Y. A., 1978, Primitives. In Shapiro (ed.), Encyclopedia of Articial
Intelligence. John Wiley and Sons, New York.
Wilson, B., 1984, Systems: Concepts, Methodologies, and Applications. John Wiley
and Sons, Chichester.
Winograd, T., 1975, Frame representations and the declarative/procedural
controversy. In Bobrow, D. G. and Collins, A. M. (eds.), Representation and
Understanding: Studies in Cognitive Science. Academic Press, New York,
185210.
Winograd, T., 1983, Language as a Cognitive Process (Volume I: Syntax).
Winograd, T. and Flores C. F., 1987, Understanding Computers and Cognition.
Wright, G. H. v., 1951, An Essay in Modal Logic. North-Holland, Amsterdam.
Wright, G. H. v., 1963, Norms and Action a Logical Enquiry. Routledge and
Kegan Paul, New York.
Yourdon, E., 1989, Modern Structured Analysis. Prentice-Hall, Englewood Clis,
New Jersey.
Yourdon, E. and Constantine, L., 1979, Structured Design. Prentice-Hall,
Yu, C. H., 1994, Abduction? Deduction? Induction? Is there a logic of
exploratory data analysis? The Annual Meeting of American Educational
Research Association, New Orleans, Louisiana,
http://seamonkey.ed.asu.edu/~behrens/asu/reports/Peirce/Logic_of_EDA.ht
ml (accessed 7 August 1998).
216 Bibliography
aordance 6162, 65
agent 46, 65
antecedent 66
authority 69
axioms 75, 134
behaviour
patterns 61
semiotic 67
substantive 45, 67
collateral analysis 43
communication 82
addressee 84, 97
preparatory conditions 86
speaker 84, 97
conceptual graphs 5456
conceptual model 50
CONTEST 180
CRIS 201
database
deductive 134
extensional 119, 134
intensional 119, 134
object-oriented 135
relational 134
semantic temporal 127, 133, 154,177
deontic logic 8994
deontic operator 93
determiner 64, 68
empirics 28
FRISCO 6
functional grammar 87
genericspecic 68
identity 68
information analysis 56, 58
information systems 2, 109
formal 110
informal 109
methodologies 5
technical 110
intention 124
invariant 62
knowledge, types 50, 52
LEGOL 7173, 143
operator 145
learning 15
MEASUR 78, 37
mode
aective 75
denotative 75
norm 98
proto- 103
specication 105
Norm Analysis 7980
NORMA 6471
Normbase 150152
object-oriented design 159, 191
object-oriented implementation 161
ontological dependency 46, 65
ontology chart 6979
organisational morphology 112
organisational onion 109
paradigm 21
objective 2124
subjective 2426
particular 61
physics 23
position
epistemological 57
ontological 57
217
Index
pragmatic information 32
pragmatics 31
problem articulation 38
proposition 85, 94
propositional attitudes 95
representation method 52
requirement analysis 60
responsibility 69, 111
analysis 103
Semantic Analysis 35, 7379, 165
semantic modelling 156
semantic primitives 50
semantic template 138
semantic unit 7576
semantics 30
semiosis 1517
semiotic diagnosis 40
semiotic framework 2635
social level 33
speech act 8386
illocutionary 84
perlocutionary 68
surrogate 140
syntactics 29
system morphology 44
time 67, 121
event time 136
transaction time 136
trigger analysis 104
unit system 43
universal 65
update 136
non-destructive 136
valuation framing 43
well-formed formula 65
218 Index

Semiotics in Information Systems Engineering

Uploaded by

Copyright:

Available Formats

Semiotics in Information Systems Engineering

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Semiotics in Information Systems Engineering

Uploaded by

Copyright:

Available Formats

This page intentionally left blank

To Lily and Jimmy

is obtained by adding an axiom:

is closed under the following rule of proof:

Propositional attitudes can be roughly indicated by two scales: the denota-

Figure 7.3. Propositional attitude scales.

You might also like