Electronic Commerce Research and Applications 6 (2007) 128–138

www.elsevier.com/locate/ecra

Using the composite act frame technique to model

‘Rules of Origin’ knowledge representations in e-government services
Wanchai Varavithya *, Vatcharaporn Esichaikul
School of Engineering and Technology, Asian Institute of Technology, P.O. Box 4, Klong Luang, Pathumthani 12120, Thailand

Available online 20 December 2006

Abstract

Rules of Origin (ROO), globally applied to determine the eligibility for trade preferences, are a vital instrument in international trade,
since they defines the country of origin of products. Whereas the technical nature of the ROO criteria is complex and voluminous, com-
mon ROO knowledge representation is a vital key in promoting interoperability and effective e-government services. This paper aims to
lay down a ROO knowledge representation model using the composite act frame technique extended from the frame-based ontology of law.
To prove the generic and extensibility aspects of the model, we conducted an assessment test with different criteria for the ROO. The
implementation of the ROO knowledge representations to support Web-based e-government services is accomplished ‘‘Rules of Origin
VERification Systems’’, known more simply as ‘‘ROVERs’’.
 2006 Elsevier B.V. All rights reserved.

Keywords: Composite act frame technique; E-government; Free trade agreements; Frame-based ontology of law; Rules of Origin

1. Introduction Major differences between the ROO and other law issues
in international trade, namely non-tariff measures2 (NTM),
With the proliferation of free trade agreements (FTAs) is that although NTM may be deployed by national gov-
in the last decade [1–4], government and international trade ernments as an indirect way to reduce the volume and value
experts have paid attention to one of its most important of trade, ROO are, however, required by the World Trade
components: Rules of Origin (ROO). ROO1 are the criteria Organization to be transparent, non-restrictive, non-dis-
needed to determine the national origin of a product [6]. torting or non-disruptive to international trade. Moreover,
ROO form a critical component of FTAs to define the con- ROO must be administered in a consistent, uniform,
ditions under which the importing economy will regard a impartial and reasonable manner [6].
product as originating in an exporting economy that As ROO criteria are difficult, onerous, and opaque for
receives preferential duty or tariff treatment from the compliance to businesses, the ROO administrative proce-
importing economy. dures create high administrative costs for the government
and businesses [8]. The motivation of our research is to
take the first step in the construction of Web-based
*
Corresponding author.
E-mail address: [email protected] (W. Varavithya).
1
The GATT Agreement defines Rules of Origin as ‘‘. . .those laws,
2
regulations and administrative determinations of general application Baldwin [7] defines non-tariff distortion as ‘‘any measure (public or
applied by any Member to determine the country of origin goods, private) that causes internationally traded goods and services, or resources
provided such Rules of Origin are not related to contractual or devoted to the production of these goods and services, to be allocated in
autonomous trade regimes leading to the granting of tariff preferences such as way as to reduce potential real world income’’. Samples of non-
going beyond the application of paragraph I of Article I of GATT 1994’’ tariff measures are export subsidies and taxes, antidumping regulations,
[5]. and controls over foreign investment.

1567-4223/$ - see front matter  2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.elerap.2006.11.005
 W. Varavithya, V. Esichaikul / Electronic Commerce Research and Applications 6 (2007) 128–138 129

ROO verification services by designing ROO knowledge Roll-up/

Absorption
representations using the composite act frame technique De minimis rule
Principle
Cumulation
extended from the frame-based ontology of law proposed
by van Kralingen et al. [9]. The significance of this Regime-wide
research is that the proposed ROO knowledge represen-
tations and ontology can be shared and reused as a com-
Product
mon ROO vocabulary, which is to say that the Specific
Substantial
meanings, properties, and relations of concepts will be Wholly Obtained/
Produced
Transformation

useful to other ROO modeling under different FTAs.

Additionally, the ROO knowledge representations con- Change in Tariff Exception
attached to Value Content Technical
tribute to the e-government interoperability, particularly Classification (CT)
particular CT Requirement

the government to businesses.

The rest of this paper is organized as follows. In Section Fig. 1. The typology of the Rules of Origin criteria. (Adapted from Garay
and De Lombaerde [10]).
2, we begin with a short description of the background of
the ROO criteria. Next, the ontology techniques and the
frame-based ontology of law are overviewed and the pro- producers of one FTA member to use non-originating
posed composite act frame technique is introduced in Sec- materials from another FTA member without losing
tion 3. In Section 4, we present the ROO knowledge the preferential status of the final product.
representations modeling based on the Thailand–Austra-  Product-specific rule: Product-specific rule is a rule that
lian FTA. In Section 5, after the composite act frame tech- is applied exclusively to a specific product under the
nique has been incorporated, the consistency and Harmonized System (H.S.).4 There are basically two
integrality of the frame-based ontology of law is investi- types of product-specific rule:
gated. Furthermore, an assessment test with other ROO (a) Wholly obtained/produced rule is conferred upon a
agreements is conducted and results are reported accord- product that is extracted, harvested, or produced in
ingly in Section 6. In Section 7, the implementation of the territory of an originate country.
the ROO knowledge representation, called the ‘‘Rules of (b) Substantial transformation rule is conferred upon a
Origin VERification Systems’’ (ROVERs), is illustrated. product of which two or more countries are involved
Finally, the discussions and conclusions of our research in the production of goods. It consists of four main
are presented in Section 8. components. First is a change in tariff classification
(CT) between the manufactured goods and the for-
2. The Rules of Origin criteria eign inputs and materials used in the production pro-
cess. Second is an exception attached to particular CT
The domain of interest in our research is the ROO that which prohibits the use of non-originating materials
is employed in FTAs and generalized systems of prefer- from a certain subheading, heading, or chapter. Third
ences3 (GSP). The basic notions, problems, and implica- is value content, which requires that the product
tions of ROO in e-government services can be discussed acquire a certain minimum local value in the export-
as follows: ing economy (or, alternatively, to remain below a cer-
tain ceiling percentage of value originating in the non-
2.1. Basic notions about the Rules of Origin criteria member economies). Fourth is the technical require-
ment, which requires the product to undergo certain
Based on a survey of various existing types of ROO [10], manufacturing operations in the original economy.
the different criteria of ROO are as follows (see Fig. 1):

 Regime-wide rule: Regime-wide rule employs a degree of 2.2. Problems related to the Rules of Origin criteria
de minimis rule, which allows for a specified maximum
percentage of non-originating materials to be used with- The ROO raise a set of problems, for instance:
out affecting origin. The roll-up or absorption principle
allows materials that have acquired origin by meeting  The complexity and voluminous nature of ROO problems.
specific processing requirements to be considered origi- ROO criteria are complex, and as a result, are difficult
nating when used as input in a subsequent transforma- and time consuming for one human being to understand
tion and the type of cumulation, which allows and verify for correctness. ROO often account for large

4
The World Customs Organization (WCO) defines the ‘‘Harmonized
Commodity Description and Coding System’’ or ‘‘Harmonized System’’ as
3
The generalized system of preferences (GSP) is a preferential tariff the nomenclature comprising the headings and subheadings and their
system extended by developed countries (also known as preference giving related numerical codes, the section, chapter and subheading notes and the
countries or donor countries) to developing countries. general rules for the interpretation of the Harmonized System [11].
130 W. Varavithya, V. Esichaikul / Electronic Commerce Research and Applications 6 (2007) 128–138

chunks of FTAs and can be very contentious. For exam- edge-based systems is the most important function of
ple, the ROO part of Thai-Australian FTA, Thai– the model where an applicant’s data is inputted through
Indian FTA, and Thai-ASEAN FTA contain more than the Website will be processed to determine the facts of
9900 rules in total. the case. The verification process is fully explained in
 ‘‘Spaghetti bowl effect’’ problem. The ROO criteria are Section 7.2.
defined in each FTA and often contain different knowl-  ROO knowledge-based systems work as a repository for
edge formulations, producing varied outcomes on the the ROO knowledge for capturing, storing, and reason-
same product. The number and disparities of ROO ing purposes. The correctness and completeness of the
across various FTAs are called the spaghetti bowl effect rules are vital to the citizen’s confidence in government
[10]. The overlapping and inconsistency of the ROO services. The authority has the responsibility of manag-
cause confusion to both the government and exporters. ing the accuracy of the ROO knowledge representations
 High administrative cost problem. The complicated and contained in the knowledge base.
difficult to understand ROO have undeniably made inter-  Human intervention facility is used mainly by an authority
national trade more costly and complex [12]. The admin- who needs to exercise judgment and discretion as speci-
istrative costs of ROO arise from the procedures required fied by the ROO criteria, namely technical requirements.
for ascertaining compliance with the requirements of the  Explanation facility is where e-government services pro-
ROO regime, especially the costs for ROO verifications vide explanations and reasons as to why the service’s
of complex and divergent ROO regimes [8]. outcomes are accepted or rejected.
 The problem of vagueness in the ROO. The ROO provide  XML Rules of Origin support e-government interopera-
guidelines for evaluating the final product and the bility by disseminating rules through a one-stop e-gov-
domestic inputs, which can be generally vague and ernment portal using XML technologies.
ambiguous leading to potential manipulation and vary-
ing interpretation on the part of authorities. Hence,
authorities and exporters have wide discretion in apply- 3. The frame-based ontology of law and the composite act
ing the ROO [8]. frame technique

We apply the frame-based ontology of law to conceptual-

2.3. The ROO verification service model ize and define elements and their relationships pertaining to
the ROO domains. The composite act frame is proposed to
In order to appreciate the contribution of the ROO construct the norm frame knowledge representations from a
knowledge representations, it is useful to understand its complex ROO argument. The review on the ontology tech-
implication in the e-government services. Fig. 2 presents niques, the concept of the frame-based ontology, and the
a conceptual model of a ROO verification service. The ser- composite act frame technique are given as follows:
vice consists of five major components as follows:
3.1. Ontology techniques
 E-government Website is where a business accesses
e-government services, submits data, and interacts with Ontology is defined by Gruber [13] as ‘‘an explicit concep-
a government agency. The ROO verification linkage tualization of a domain’’, which describes the entities and
between the e-government Website and the ROO knowl- relations taken to exist in the domain. Jasper and Uschold

Interoperate with

XML Rules of
Disseminate Transform
Origin
One Stop
E-Government Portal

Rules of Origin
E-Government
Interact Verify Knowledge-based Manage
Website
Systems

Send
Explain results results Specify rules
Business
Decision
Send Human making
Explanation Facility
results Intervention Facility

Authority

Fig. 2. The conceptual model of the Rules of Origin verification service.

 W. Varavithya, V. Esichaikul / Electronic Commerce Research and Applications 6 (2007) 128–138 131

[14] identify four main categories of ontology applications, Table 1

namely (i) neutral authoring is an information artifact that Norm frame in the context of the ROO domains
is authored in a single language and is converted into a dif- Element Descriptions
ferent form, (ii) ontology as specification is created and used Norm identifier The tariff number of product under the H.S. is
as a basis for the specification and development of some assigned as a point of reference, which could be
software, (iii) common access to information is required by chapters, headings, or subheadings
Promulgation The source of the norm which is the name of an
one or more persons or computer applications, and is agreement
expressed using unfamiliar vocabulary, or is in an inaccessi- Scope The range of applications is divided into a regime-
ble format, and (iv) ontology-based search is used for search- wide rule or a product-specific rule
ing an information repository for desired resources. Conditions The ROO verification result is assigned to *True*
The design of ROO knowledge representations in this Act identifier The reference to separate a single or a composite
act frames
research focuses on a common access to ROO knowledge
either through a human or a computer application. We uti-
lize the top-down method [15] where technicians use an Table 2
existing ontology and specify or generalize it to create Concept frame in the context of the ROO domains
another one. As legal ontology has been extensively Element Descriptions
researched in the field of artificial intelligence (AI) and laws Concept Concept pertaining to the ROO agreement
[16], it can serve as an excellent tool to deal with a complex Type Term or mathematics
structure of legal entities that govern e-government Promulgation The name of an agreement
applications. Scope Regime-wide rule or product-specific rule
Conditions Text descriptions or mathematic formulae
3.2. Frame-based ontology of law and the Rules of Origin
domain
 The act frame represents the dynamic functional aspects,
The ontology-based conceptualization of knowledge which affect changes in the state of the world. We desig-
representation from knowledge formulation in the legal nate the ‘‘act identifier element’’ as a function that
domain has been performed in several research works, expresses a behavior of rules and encapsulates operation
e.g. McCarty’s language for legal discourse [17], Stamper’s details under the ‘‘manner’’ element. Each act frame
NORMA formalism [18], Valente’s functional ontology of must contain only a single function. The ‘‘means’’ ele-
law [19], and van Kralingen’s frame-based ontology [9]. ment describes the knowledge formulations of a specific
Because of the generic aspect and the incorporation of function. The ‘‘agent’’ element describes input variables
the first-order formulation into the element of knowledge into functions while the ‘‘final state’’ or output must be
representations of van Kralingen’s frame-based ontology, ‘‘true’’. The norm frame and the act frame link with each
this ontology is suited to conceptualizing and defining ele- other through the element ‘‘act identifier’’. Table 3
ments contained in the ROO domains. Van Kralingen’s shows a complete act frame element description. It
frame-based ontology is comprised of three frame should be highlighted that we categorize the ‘‘act type’’
structures: the norm frame, the act frame, and the con- element of the act frame into two separate decision-mak-
cept-description frame. Each frame further comprises a ing actions: logic and human intervention.
vocabulary or an element necessary to instantiate the frame (a) Logic is used for routine decision-making, which can
structures. Descriptions of each frame in the context of the be made by verifying structured data against the
ROO criteria are as follows: applicable rules.
(b) Human intervention is applied to decisions involved
 The norm frame is a set of general rules, standards, and in vagueness problems, namely technical require-
principles of behavior with which subjects of laws are ments. With this act type, a knowledge base can indi-
bound to comply. This is the most important element cate whether the ROO verification process needs
of the legal systems, as it puts together every important human involvement in the decision-making.
element necessary for reasoning purposes. Under the
ROO context, we designate the elements of norm as Table 3
shown in Table 1. Act frame in the context of the ROO domains
 The concept frame deals with the meaning of concepts Element Descriptions
found in the domain. The definition of concepts or fac-
Act identifier Function ()
tors pertaining to the FTA must be clearly specified in Agent Input variables
the element ‘‘condition’’, which could be in the format Scope Regime-wide rule or product-specific rule
of descriptive text or mathematical formulae. The norm Act type Logic or human intervention
frame and the concept frame link through the element Means Description of functions
‘‘promulgation’’. Table 2 shows the concept frame in Manner First order logic formulae
Final state Output *True*
relation to the ROO description.
132 W. Varavithya, V. Esichaikul / Electronic Commerce Research and Applications 6 (2007) 128–138

Norm Frame
Norm identifier Concept Frame
Promulgation Promulgation
Act identifier: Composite Act Frame
Use FOL connective:
Conjunction, Disjunction

Act Frame Act Frame Act Frame

Act identifier Act identifier Act identifier

Fig. 3. Relationship between the norm frame, the concept frame, and the composite act frames.

3.3. The composite act frame technique m () = Imported material

Chapter () = Classified in chapter: the first two
Van Kralingen’s frame-based approach keeps the ele- digits of the H.S.
ment ‘‘act identifier’’ as a reference link between the norm Heading () = Classified in heading: the first
frame and the act frame. However, in order to deal with a four digits of the H.S.
complex semantic unit of ROO, we propose the composite SubHeading () = Classified in subheading: the
act frame technique similar to a concept of a composition first six digits of the H.S.
of functions in discrete mathematical concept [20] to relate Rvc () = Regional Value Content in percentage
one or more act frames using a connective relationship Arvc () = Regional Value Content Agreement
from the language of first-order logic (FOL), namely con- in percentage
junction () and disjunction (). Ivc () = Imported Value Content in percentage
To instantiate the norm frame of each ROO of each prod-  Function symbols
uct, the act identifier element of the norm frame could be rep- InRangeOf (ti, tj) = Range of Ætariff number
resented in a single act frame or a composite act frame, lower limitæ and Æ tariff number upper limitæ
depending on the complexity of the knowledge formulation under the H.S.
of each rule. Fig. 3 depicts the relationship model between Step 2: Instantiate the concept frame and act frame
the norm frame, the concept frame, and the composite act  Instantiate the concept frame
frames. The concept frame is constructed to describe
ROO terms or concepts pertaining to FTAs,
4. Modeling the rules of origin knowledge representations such as the de minimis content rule, the regional
value content, etc. An excerpted sample of a
We select the Thailand-Australian FTA [2], which was concept frame instantiated from the Thai-Aus-
entered into force on January 1, 2005, as a point of depar- tralian FTA documents is shown in Table 4.
ture. The reason behind this choice is the volume and com-  Instantiate the act frame
plexity of the rules contained in the agreement. We expect The act frame is used as a separate function for
the model gleaned from the agreement to be generic the representation of rule behavior. The
enough to be applied or extended to ROO criteria in other excerpted sample of the act frame is shown in
FTAs. Our modeling methodology consists of the follow- Table 5. The functions contained in each act
ing three steps. frame are very important in the instantiate act
identifier of the norm frame for each product.
Step 1: Designate a fundamental structure of the ROO In order to instantiate the act frame from a
descriptions. voluminous ROO criterion, we use the common
The knowledge formulations of the ROO criteria rule approach to analyze and assemble similar
are similar to legal arguments, as they express rule product-specific rule knowledge formulation
statements stipulating that a product under con- patterns. We designate an act identifier’s name
sideration ought to or ought not to originate in and construct a manner element in the language
the country of origin. In order to enclose the of first-order logic as shown in Table 6.
first-order formulation in the frame-based ontol- Step 3: Use the composite act frame technique to instanti-
ogy, we have designated the object’s names, pred- ate the norm frame.
icate symbols, and function symbols pertaining to The norm frame in the context of the ROO con-
the ROO domain as follows. veys information to fulfill its function for the ver-
 Object names ification of a product in question. Therefore, the
t = Tariff number under the H.S. ROO norm frame serves as a scheme of rule inter-
 Predicate symbols pretation classified under the product H.S. For a
p () = Exported product product with one ROO criteria, using a single act
 W. Varavithya, V. Esichaikul / Electronic Commerce Research and Applications 6 (2007) 128–138 133

Table 4
The sample of a ‘‘regional value content’’ concept frame

{SubHeading (m(t)) 2 InRangeOf (SubHeading (t)i, SubHeading (t)j)}

{(Chapter (m(t))  Heading (m(t))  SubHeading (m(t))) 62 InRangeOf

6 InRangeOf
Element Descriptions
Concept Regional value content
Type Mathematics

{Heading (m(t)) 2 InRangeOf (Heading (t)i, Heading (t)j)}

Promulgation Thai-Australian FTA

{(Chapter (m(t))  Heading (m(t))  SubHeading (m(t))) 2

Scope Product-specific rule

Manner elements in the language of first-order logic

Conditions Regional value content (%) = (FOB  VNM/FOB) * 100
* FOB = Free on Board; VNM = Value of all Non-originating Materials.

{Chapter (t)  Heading (t)  SubHeading (t)}

Table 5

{SubHeading (m(t)) 6¼ SubHeading (t)}

{SubHeading (m(t)) = SubHeading (t)}

The sample of a ‘‘change in tariff classification criteria’’ act frame

(SubHeading (t)i, SubHeading (t)j)}

Element Descriptions

{Heading (m(t)) 6¼ Heading (t)}

{Heading (m(t)) = Heading (t)}

{Chapter (m(t)) 6¼ Chapter (t)}

{Rvc(*##%*)PArvc (*##%*)}
Act identifier @CTOC (*Chapter (t)*)

(Heading (t)i, Heading (t)j)}

Agent {Heading (p(t))  SubHeading (p(t))}, {Chapter (m(t))}
Scope Product-specific rule
Act Type Logic

{Ivc (*##%*) = *0*}

Means Change to (heading ’####’, subheading ’######’) from
any other chapter
Manner {Chapter (m(t)) 6¼ Chapter (t)}

The Rules of Origin act frame knowledge representations

Final state *True*

frame is sufficient. However, for a product that

consists of complex ROO criteria, applying the
composite act frame technique with the ROO

@CFSR (*SubHeading (t)i, SubHeading

@CTSR (*SubHeading (t)i, SubHeading

@CFHR (*Heading (t)i, Heading (t)j*)

@CTHR (*Heading (t)i, Heading (t)j*)

act frame knowledge representations from Table
6 can be very useful. To articulate the general

@TECH (*Chapter (t)  Heading

application of the composite act frame technique

@RGVC (*P*,*Arvc(*##*)*)
@CTOS (*SubHeading (t)*)

@CFTS (*SubHeading (t)*)

in the instantiation of the norm frame from com- @CTOH (*Heading (t)*)

@CFTH (*Heading (t)*)

(t)*)

plex ROO criteria, two examples of products HS.

(t)  SubHeading (t)*)

Act identifier element

620500 and H.S. 848220 with the norm frame (see

(*Chapter

Tables 7 and 8) and the composite act frame (see

@WHOL(*0*)

Figs. 4 and 5) are illustrated.

@CTOC

Example 1: Product HS 620500 (men’s or boy’s shirts)

The product-specific Rules of Origin act frame knowledge representations pattern

(t)j*)

(t)j*)
with ROO ‘‘Change to heading 6205 from any other
chapter, provided that the good is both cut and sewn or

(6) Regional content value not less than ‘##’ % of FOB value of the good.
otherwise assembled in the territory of one or both of
the parties and there is a regional value content of not less
(3) Change to (Heading, Subheading) from (Heading, Subheading).

than 55%’’.
(2) Change to (Chapter, Heading, Subheading) from any other

(4) Change to (Heading, Subheading) from range of (Heading,

(5) Change to (Chapter, heading, subheading) except range of

Example 2: Product HS: 848220 (tapered roller bearings,

including cone and tapered roller assemblies) with ROO
‘‘A change to subheading 848220 from any subheading
The Rules of Origin knowledge formulation patterns

except from subheadings 848210 through 848280 and inner

or outer rings or races of subheading 848299; OR A change
to subheading from inner or outer rings or races of sub-
heading 848299, whether or not there is also a change from
any subheading outside that group provided there is a
(Chapter, Heading, Subheading).

(Chapter, Heading, Subheading).

regional value content of 45%’’.

5. Consistency and integrality of the frame-based ontology of

(7) Technical requirements

law after adding the composite act frame technique

(1) Wholly obtained

The consistency and integrality of the frame-based

Subheading).

ontology of law after the composite act frame technique

has been added is a critical issue that needs examination.
Table 6

Based on Gomez-Perez’s [21] definition of consistency

and completeness of ontology, a qualitative measure is
134 W. Varavithya, V. Esichaikul / Electronic Commerce Research and Applications 6 (2007) 128–138

Table 7 of ROO knowledge formulations in the ROO domains.

Norm frame knowledge representation of H.S. 620500 Elements of the composite act frame represented in the
Element Value frame-based ontology are expressed as the same knowledge
Norm Identifier 620500 intended in the ROO knowledge formulations. As such, the
Promulgation Thai-Australian FTA composite act frame technique clearly promotes a better
Scope Product-specific rule flexibility and extendibility as an integral part of the origi-
Conditions *True*
nal frame-based ontology of law.
Act identifier {@CTOC(*62*)  TECH(*6205*) 
@RGVC(*P*,*55*)}
6. Assessment test

Table 8 To prove the generic and extendibility aspects of

Norm frame knowledge representation of H.S. 848220 the ROO knowledge representation, we conducted an
Element Value assessment test with other ROO criteria besides the Thai-
Norm identifier 848220 Australian FTA [2], namely the Thai–Indian FTA [3], the
Promulgation Thai-Australian FTA Thai-ASEAN FTA [1], and the Thai-European Union
Scope Product-specific rule (EU) GSP [22].
Conditions True On the generic modeling aspect, after applying the ROO
Act identifier {(@CTOS(*848220*) 
knowledge representations to instantiate norm frames per-
@CTSR (*848210*,*848280*) 
TECH(*848299*))  (TECH(*848299*)  taining to the three other agreements, we found that Indian
@RGVC(*P*,*45*))} and ASEAN ROO knowledge formulations can completely
be instantiated without a need to create a new act frame.
However, for EU ROO criteria, we found that there were
applied with the focus on the structural view of knowledge some variances in knowledge formulations that could not
and a semantic definition. This involves feedback from be instantiated from existing knowledge representations
ROO experts on the semantic correctness between ROO (see Table 9).
knowledge formulation and knowledge representation after To test the extendibility aspect, after analyzing the dis-
using the composite act frame technique. A review of the tinctive ROO criteria in the Thai-EU GSP agreement, we
consistency by domain experts reveals that by adding the encountered two new knowledge formulation patterns,
composite act frame technique, the formal definition of which were different from existing act frame knowledge
norm, act and concept of the frame-based ontology are representations. As such, we have extended additional act
consistent. This means that the meaning of each element frames as shown in Table 10. The number of new act
in each frame is consistent with the ROO world. On the frames used in the instantiation of norm frame results in
analysis of completeness, the composite act frame tech- 5 single act frames of @EXEP, 185 composite act frames
nique helps domain experts extend and cover a wide range of @EXEP, and 9 composite act frames of @NOGO.

Norm Frame
Norm identifier 620500

Act identifier {@CTOC(*62*) ∧ TECH(*6205*) ∧ @RGVC(*≥ *,*55*)}

Act Frame Act Frame Act Frame

@CTOH (Chapter (t)) @TECH (Heading ( t )) @RGVC (*>=* , (Arvc ())

Fig. 4. The composite act frame diagram of H.S. 620500.

Norm Frame
Norm identifier 848220
{(@CTOS(*848220*) ∧ @CTSR(*848210*,*848280*) ∧
Act idengifier
TECH(*848299*)) ∨ (TECH(*848299*) ∧ @RGVC(*≥*,*45*))}

Act Frame Act Frame Act Frame Act Frame

@CTSR (SubHeading ( t )i), @TECH (Chapter (t ) V Heading (t ) @RGVC (*>=* , (Arvc ())
@CTOS (SubHeading ( t ))
SubHeading (t )j) V SubHeading ( t ))

Fig. 5. The composite act frame diagram of H.S. 848220.

 W. Varavithya, V. Esichaikul / Electronic Commerce Research and Applications 6 (2007) 128–138 135

Table 9
Number of single act frame (SAF) and a composite act frame (CAF) from different agreements
Act identifier Australia India ASEAN EU
SAF CAF SAF CAF SAF CAF SAF CAF
@WHOL () – – 12 – – – 100 –
@ CTOC () 196 185 – – – – 1 14
@ CTOH () 623 455 – 51 – – 323 361
@ CTOS () 1158 367 – 17 – – – –
@ CFTH () – 18 – – – – 8 5
@ CFTS () – 100 – – – – – –
@ CFHR () – 1 – – – – – 1
@ CFSR () – 14 – – – – – –
@ CTHR () – 50 – – – – 3 1
@ CTSR () – 121 – – – – – –
@ RGVC () 1 442 4 68 6523 – 128 381
@ TECH () 38 151 – – – – 322 85
N/A – – – 190

Table 10
Extended act frame knowledge representations
Knowledge formulation patterns Act identifier element Manner elements
(1) Import value content of material under the H.S. does @EXEP (*Heading m(t)*, {Ivc (m(t)(*##%*) P Aexp
not exceed *##* % of the ex-works price of the product Aexp (*##*)) (*##%*)}
(2) The value of all the non-originating materials used does @NOGO (*0*) {Rvc (*##%*) P Ivc
not exceed the value of all the originating materials used. (*##%*)}
*Aexp () = Agreement ex-works price of the product in percentage.

7. Implementation based systems’’ is the ROO verification function. Fig. 7 pre-

sents a verification process algorithm that verifies the ROO
We have implemented the ROO knowledge representa- information submitted from citizens through the E-govern-
tion model in the actual e-government services named the ment Website. At the beginning, the E-government Website
‘‘Rules of Origin VERification Systems’’: ROVERs, oper- transmits the ROO information which is comprised of
ated by Ministry of Commerce, Thailand [23]. ROVERs product tariff number (p(t)), agreement code (a(c)), and Bill
has been providing on-line Rules of Origin verification ser- of Material information (BOM) to the ROO knowledge
vices to Thai businesses since January 2005. The basic goal based systems for verification purposes. p(t) and a(c) are
of ROVERs is to answer whether a product under consider- used as indices to retrieve the corresponding act identifier
ation originated in Thailand in connection with applicable (ActID) in a string format.
ROO criteria under specific trade agreements. To manage The verification process is composed of two steps: cre-
ROO knowledge and verify ROO information, a knowledge ating an act identifier element object and evaluating the
editor and a verification process are presented as follows: ROO information. The purpose of the create act identifier
element object function is to instantiate the ActID to an
7.1. Rules of Origin knowledge editor act identifier element object ActIDElement (e.g.
@CTOC(t)). ActIDElementobject encapsulates Manner
The product-specific rule knowledge editor interface is elements (e.g. {Chapter (m(t)) 6¼ Chapter (t)}) for reason-
designed for domain experts to manage rules by viewing ing purposes. IfActID is a composite act frame, the
existing ROO knowledge formulations and constructing Iterator pattern technique [24] is used to recursively sub-
or editing a product-specific knowledge representation for string composite ActID string to a left operand, an oper-
each product using a single or composite act frame tech- ator name, and a right operand and instantiate each
nique (see Fig. 6). To ensure the correctness of the compos- ActID string to ActIDElement objects. To compose ActI-
ite act frame, the simulate feature is provided for domain DElement objects into tree structures that represent
experts to enter essential bills of material data and verify logical hierarchies of the rule objects, the Composite pat-
results accordingly. tern technique [24] is applied. In the evaluation of the
ROO information against the rules, p(t) and BOM will
7.2. Rules of Origin verification process be used as input variables to Act identifier elements and
execute a logical operator in order to obtain verification
As shown in Fig. 2, the most important link between the results which are accepted, rejected, or result in human
‘‘E-government Website’’ and ‘‘Rules of Origin knowledge- intervention.
136 W. Varavithya, V. Esichaikul / Electronic Commerce Research and Applications 6 (2007) 128–138

Fig. 6. Product-specific rule knowledge editor user interface.

8. Discussions and conclusions technologies. The Semantic Web is based on machine-pro-

cessable semantics of data [26]. It can be regarded as
The main contribution of this paper is to lay down the grounding technologies to enable ROO information pro-
ROO knowledge representation model using the composite cessing, collaboration, and workflows for the e-government
act frame technique. The reflections from our research can interoperability. Despite the fact that electronic data inter-
be discussed as follows: change (EDI) has been able to increase the efficiency and
effectiveness of government services in Thailand, EDI is
8.1. Advantage of the composite act frame technique based on the transfer of fixed transaction sets and business
rules which are extremely difficult to deal with in the chang-
To demonstrate a significant advantage of the composite ing pace of the current trade regulation environment [27].
act frame technique over other techniques, the ontology of Additionally, EDI is an isolated technology of which inter-
law proposed by Valente et al. [19] is chosen for compari- changed data cannot be easily integrated with other busi-
son in terms of reusability and extendibility. We found that ness applications [28].
the Valente’s legal ontology is also an effective technique As concerns the implementation of ROO verification
for modeling the knowledge pertaining to the ROO world. services, event-condition-action (ECA) rules can be consid-
However, the Valente’s legal ontology considers actions as ered as a prime candidate technology to realize the seman-
part of world knowledge but they do not constitute a sep- tic Web technologies as long as rules are not embedded in
arate class in the definition of world knowledge [25]. As programming codes. ECA for the semantic Web [29] can
such, it is not helpful to reuse existing actions for other support interoperability among rule-using applications that
rules or extend a new rule to disentangle the complexity make use of heterogeneous members of different rule sys-
and heterogeneous nature of the ROO. The composite tems. Adapting ROO knowledge representations and
act frame technique can be leveraged as a powerful tech- ontology to ECA deserves further study in the future.
nique to model, reuse, and extend to new acts in order to
deal with the complexity of the ROO. 8.3. E-government service to businesses

8.2. E-government interoperability The standardization of organized and semantically rich

ROO knowledge representations and ontology is very
Besides using ROO knowledge-based systems to provide important in promoting government-to-business (G2B) or
e-government web-based services to citizens over a net- business-to-government (B2G) services. To enable the busi-
work, the effort in the design of ROO ontology also aims ness companies to enact with the Semantic Web technolo-
to support interoperability among government and busi- gies, the government has a role in maintaining a single
ness applications and platforms using the Semantic Web point of information about the ROO knowledge model
 W. Varavithya, V. Esichaikul / Electronic Commerce Research and Applications 6 (2007) 128–138 137

Fig. 7. Rules of Origin verification processes algorithm.

and verification processes. Businesses may apply the ROO. However, as the FTAs in international trade are
semantic matching technique [30] as a key component to expected to multiply in the future, a new kind of ROO cri-
tailor ROO forms to the specific needs of business applica- teria will certainly emerge. As such, the dissimilarity and
tions and platforms. With the business semantic applica- reusability of ROO knowledge warrant further studies in
tions, the format of the trade’s related documents and the the future.
ROO information are automatically generated and pre- In conclusion, this paper demonstrates the implications
sented as required by each FTA. of the composite act frame technique and legal frame-based
ontology in the design of ROO knowledge representations
8.4. Rules of Origin knowledge management and effective e-government services. Nevertheless, the com-
posite act frame technique proposed in this paper can be
As its technical nature makes the ROO difficult and the extended as a solution to other non-law areas of e-govern-
variations of its standards across FTAs have made it more ment. For example, for e-government citizen services that
perplexing to exporters, the government must encourage require integrated functions from various agencies in a
citizens to access and utilize ROO knowledge effectively seamless manner, the composite act frame can be general-
in order to maximize the benefits from FTAs. With effec- ized in the design of citizen assistance by defining a partic-
tive ROO knowledge management, businesses are ular service as an act identifier element in the act frame and
expected to collect and retain ROO knowledge in order constructing a logical composite services (‘‘act identifier ele-
to plan and produce products based on the applicable ment’’) in the norm frame based on a citizen’s individual-
138 W. Varavithya, V. Esichaikul / Electronic Commerce Research and Applications 6 (2007) 128–138

ized situation. General citizen situations and government Workshop on Knowledge Acquisition, Modeling and Management
services can be defined in the concept frame. Lastly, we Voyager Inn, Canada, 1999.
[15] G. Lame, Constructing an IR-oriented legal ontology, in: Proceedings
must reiterate that the implications of knowledge from of Second International Workshop on LEGAL ONTOLOGIES
AI and legal domains in e-government still remain a huge JURIX 2001: the 14th Annual International Conference on Legal
opportunity for researchers to explore, not only to support Knowledge and Information Systems, 2001, pp. 31–36.
decision-makers in a single public agency but also to pro- [16] P. Visser, T. Bench-Capon, The formal specification of a legal
mote an integrated e-government and good governance ontology, in: Proceedings of JURIX’96 Legal Knowledge Based
Systems: Foundations of legal knowledge systems, 1996, pp. 15–24.
to society as a whole. [17] L.T. McCarty, A language for legal discourse, I. basic features, in:
Proceedings of the Second International Conference on Artificial
Intelligence and Law, ACM, Vancouver, Canada, 1989, pp. 180–189.
References [18] R.K. Stamper, Signs, information, norms and systems, in: B.
Holmqvist, P.B. Anderson (Eds.), Signs at Work: Semiotics and
[1] Association of Southeast Asian Nations, ASEAN Free Trade Area, Information Processing in Organisation, De Bruyter, Berlin, Ger-
2005. www.aseansec.org/12025.htm (last accessed October 30, 2006). many, 1996, pp. 349–397.
[2] Australian government, Trade Policy, 2005. www.dfat.gov.au/trade/ [19] A. Valente, J. Breuker, B. Brouwer, Legal modeling and automated
(last accessed October 30, 2006). reasoning with ON-LINE, International Journal of Human-Com-
[3] India Thai Chamber of Commerce, Thai–India Free Trade Agree- puter Studies (51) (1999) 1079–1125.
ment, 2005. www.itcc.or.th/default.asp (last accessed October 30, [20] J.E. Munro, Discrete Mathematics for Computing, Chapman & Hall,
2006). 1992.
[4] Thai government, Free Trade Agreement, 2005. www.thaifta.com [21] A. Gomez-Perez, Some ideas and examples to evaluate ontologies, in:
(last accessed October 30, 2006). Proceedings of Artificial Intelligence for Applications, 11th Confer-
[5] Agreement on Rules of Origin, Uruguay Round Agreement, 2006. ence, 1995, pp. 299–305.
www.wto.org/English/docs_e/legal_e/22-roo_e.htm (last accessed [22] European Union, Generalised System of Preferences, 2005. euro-
October 30, 2006). pa.eu.int/comm/trade/issues/global/gsp/pr211205_en.htm (last
[6] J.H.H. Weiler, S. Cho, International and regional trade law: the law accessed October 30, 2006).
of the World Trade Organization, NYU School of Law, 2004. [23] The Rules of Origin verification services, Ministry of Commerce,
[7] R. Baldwin, Non-tariff Distortions in International Trade, Brookings Thailand, 2005. 61.19.224.73/home/main_homefirst.aspx (last
Institution, Washington DC, 1970. accessed October 30, 2006).
[8] A. Estevadeordal, K. Suominen, Measuring Rules of Origin in the [24] E. Gamma, R. Helm, R. Johnson, J. Vlissides, Design Patterns and
World Trading system and proposals for multilateral harmonization, Elements of Reusable Object-oriented Software, Addison-Wesley,
in: APEC Capacity-Building Workshop on Quantitative Methods for Boston, MA, 1995.
Assessing NTMs and Trade Facilitation, 2003. [25] P.R.S. Visser, T.J.M Bench-Capon, A comparison of two legal
[9] R.W. van Kralingen, P.R.S. Visser, T.J.M. Bench-Capon, H.J.V.D. ontologies, in: Proceedings of the 1st Workshop on Ontologies in
Herik, A principled approach to developing legal knowledge systems, Law, LEGONT-97, Melbourne, Australia, 1977, pp. 37–45.
International Journal of Human-Computer Studies (51) (1999) 1127– [26] T. Berners-Lee, J. Hendler, O. Lassila, The Semantic Web, Scientific
1154. American, 2001.
[10] L.J. Garay, P. De Lombaerde, Preferential rules of origin: models and [27] W. Hasseibring, H. Weigand, Languages for electronic business
levels of rulemaking, in: UNU-CRIS/LSE Workshop on The Inter- communication: state of the art, Industrial Management and Data
action between Levels of Rulemaking in International Trade and Systems 101 (5) (2001) 217–226.
Investment, 2004. [28] G. Antoniou, F. van Harmelen, A Semantic Web Primer, Massachu-
[11] The World Customs Organization, Harmonized System, 2006. setts Institute of Technology, Cambridge, MA, 2004.
www.wcoomd.org/ (last accessed October 30, 2006). [29] G. Papamarkos, A. Poulovassilis, P.T. Wood, An event-condition-
[12] World Trade Organization, The changing landscape of RTAs, in: action language for XML, in: Proceedings of the WWW 2002
Seminar on RTA and the WTO, WTO Secretariat, Geneva, 2003. Conference, Hawaii, 2002, pp. 486–495.
[13] T. Gruber, A translation approach to portable ontologies, Knowledge [30] F. Giunchiglia, M. Yatskevich, E. Giunchiglia, Efficient semantic
Acquisition 5 (2) (1993) 199–220. matching, in: Proceedings of The Semantic Web: Research and
[14] R. Jasper, M. Uschold, A framework for understanding and Applications: Second European Semantic Web Conference, ESWC
classifying ontology applications, in: Proceedings of the Twelfth 2005, Greece, 2005, pp. 272–289.