lOMoARcPSD|15883110

KRR - UNIT 4 (Part-1) - Lecture notes

Knowledge representation and reasoning (Shadan College of Engineering and

Technology)

Scan to open on Studocu

Studocu is not sponsored or endorsed by any college or university

Downloaded by g.sirisha Siri ([email protected])
 lOMoARcPSD|15883110

KNOWLEDGE REPRESENTATION AND REASONING

UNIT - IV

Processes: Times, Events and Situations, Classification of processes, Procedures, Processes and
Histories, Concurrent processes, Computation, Constraint satisfaction,
Change Contexts: Syntax of contexts, Semantics of contexts, First-order reasoning in contexts,
Modal reasoning in contexts, Encapsulating objects in contexts.

4. 1 Times, Events, and Situations

Charles Sanders Peirce was the first modern logician to treat processes and events as entities
distinct from the things that participate in the processes.
Another pioneer in logic, Alfred North Whitehead (1929), went further in making processes the
primary entities in his ontology.
Yet elementary logic books relegate processes to a secondary starus in comparison with the more
stable things called objects. A typical logic book would map a sentence like Brutus stabbed
Caesar to the formula
stabbed(Brurus, Caesar).

This translation ignores the detail implicit in the English sentence. It would require different
predicates for present, past, and future tenses. it does not allow further relations to be linked to
the verb, such as an adverb violently or a prepositional phrase with a shiny knife; and it cannot
support cross references from other sentences,such as The stabb'ing was violent.

Terence Parsons (1990) used event variables to represent the sentence Brutus stabbed Caesar:

The first line of the formula says that there exists an interval I, where I is before now.
The second line says that there exists a time t, which is contained in 1
The third line says that there exists a stabbing event e, the agent of e is Brutus, the patient
of e is Caesar, and e culminates at time t.
It captures more of the meaning in the English sentence.

SITUATIONS AND EVENTS. With his correlational nets, Silvio Ceccato (1961) was one of
the first computational linguists to map nouns, verbs, and adjectives to nodes with equivalent
status.

Figure 4. 1 shows a conceptual graph for the sentence Brutus stabbed Caesar violently with a
shiny knife. Figure 4. 1 incorporates some abbreviations that simplify the notation. The Past
relation, which is not a primitive, is defined by the following lambda expression:

Downloaded by g.sirisha Siri ([email protected])

 lOMoARcPSD|15883110

When the abbreviations for situations and tenses are expanded, Figure 4. 1 may be written in the
following form:

MULTIPLE OCCURRENCES.
The Past relation in Figure 4. 1 , which is attached to a concept of type Situation, shows that
everything described inside the box occurred at some time in the past. That relation could also be
attached directly to the verb, where it would indicate that the event expressed by the verb
occurred in the past without explicitly saying anything about the time of the other entities. As
an example, the sentence Yojo chased a mouse could be translated to a CG with the Past relation
attached to the concept [Chase] :

To show that that all three entities occurred together in the same situation, the Past relation can
be attached to a concept of type Situation, which encloses the concepts of the act and its
participants:

Downloaded by g.sirisha Siri ([email protected])

 lOMoARcPSD|15883110

ACTIONS AS ROLES:
Some verbs directly describe an action's form or structure, but others describe an action by its
role. Consider the sentence Mary hid the ball by placing it in a box. Hiding and placing are two
different ways of describing exactly the same action. An external observer could see Mary
"placing" the ball by the form of her action, but could not tell that she was "hiding" it without
knowing her intentions. As another example, consider the next two sentences:

4.2 Classification of Processes

Processes can be described by their starting and stopping points and by the kinds of changes that
take place in between. Figure 4.3 shows the category Process subdivided by the distinction of
continuous change versus discrete change.
In a continuous process, which is the normal kind of physical process, incremental changes take
place continuously.
In a discrete process, which is typical of computer programs or idealized approximations to
physical processes.
changes occur in discrete steps called events, which are interleaved with periods of inactivity
called states. A continuous process with an explicit starting point is called an initiation; one with
an ending point is a cessation; and one whose endpoints are not being considered is a
continuation.

At the leaves of Figure 4.3 is an icon that illustrates the kind of change:
A vertical bar indicates an endpoint, a wavy horizontal line indicates change, and a straight
horizontal line indicates no change.
A discrete process is a sequence of states and events that may be symbolized by a chain of
straight and wavy lines separated by vertical bars.
A continuous process may be symbolized by a continuous wavy line with occasional vertical
bars. Those bars do not indicate a break in the physical process, but a discontinuity in the way
the process is classified or described.

Downloaded by g.sirisha Siri ([email protected])

 lOMoARcPSD|15883110

The categories in Figure 4.3 could be differentiated further by considering an agent's intentions.
A cessation that satisfies the agent's goals is a success, one that does not satisfy the goals is a
failure.

4.3 Procedures, Processes, and Histories:

Petri nets, designed by Carl Adam Petri (1962), are a generalization of state-transition diagrams
for representing concurrent processes. For object-oriented design, Petri nets have been adopted
as the basis for activity diagrams in the Unified Modeling language (UML).
Figure 4.4 shows how a flow chart and a finite-state machine can be merged to form a Petri net,
which is more general than either of them. The circles of the Petri net, which are called places,
correspond to the states of the finite-state machine; the bars, called transitions, correspond to the
events of the flow chart.
- Places (circles): Representing resources, conditions, or states
- Transitions (bars): Representing events, actions, or changes
- Arcs (arrows): Connecting places and transitions, indicating flow of resources or control

Downloaded by g.sirisha Siri ([email protected])

 lOMoARcPSD|15883110

Figure 4.5 shows the first transition of the Petri net and its translation to a conceptual graph.
To distinguish processes, procedures, and histories, some further distinctions must be observed:
 A process is an evolving sequence of states and events, in which one of the states or
events is marked current at a context-dependent time called #now.
 A procedure is a pattern or script that determines the types of states and events that may
occur in an entire family of processes. Each process in the family is called an activation
of the procedure.
 A history is a record of the sequence of states and events that existed in the evolution of
some process. Each state and event of a history may be marked with a time stamp that
records its point in time or duration.

Downloaded by g.sirisha Siri ([email protected])

 lOMoARcPSD|15883110

4.4 Concurrent Processes

Although flow charts and finite-state machines can represent branches and loops, they are limited
to sequential processes. The major strength of Petri nets is their ability to represent parallel or
concurrent processes.
To illustrate the flow of tokens, Figure 4.7 shows a Petri net for a bus stop where three tokens
represent people waiting and one token represents an arriving bus.

Petri Nets are used to:

- Model concurrent and parallel processes

- Analyze system behavior and performance
- Validate and verify system properties
- Visualize complex systems and relationships

In Knowledge Representation, Petri Nets are used to:

- Represent knowledge about dynamic systems

- Model reasoning and problem-solving processes
- Integrate knowledge from multiple sources
- Facilitate decision-making and planning

Petri Nets provide a powerful and flexible framework for modeling and analyzing complex
systems, making them a valuable tool in Knowledge Representation.

THEOREMS ABOUT PETRI NETS. The places and transitions of a Petri net form a static
pattern of types; the tokens flowing through the net form a dynamic pattern of instances. From
the pattern of types, it is possible to prove theorems that predict the permissible changes in the
instances:
 If a transition has n input places and m output places, then every time it fires,there is a net
gain of (m- n) tokens on the Petri net.
 If for every transition, the number of inputs equals the number of outputs, then the
number of tokens on the Petri net remains constant.

Downloaded by g.sirisha Siri ([email protected])

 lOMoARcPSD|15883110

 On any cycle, if every place has one input arc and one output arc that is part of the cycle,
then the number of tokens on the cycle remains constant (even though the number of
tokens on the net as a whole may be changing).

4.5 Computation:

As purely declarative notations, logic and mathematics have no inherent directionality. A

computation, however, always has a purpose. It must thread its way through a tangle of
relationships to find the answer to some question. As an example, Newton's famous equation
relates the force F on a body to its mass m and acceleration a:

F= m a

By the way it's written, this equation suggests that force is the unknown result to be computed
from the given mass and acceleration. Yet the equation could just aswell be used to compute the
mass from F and a or the acceleration from F and m.
Similar observations apply to logical implications, which can be used in forward chaining
or backward-chaining applications. The following implication as a rule for the harmonious
resolution of a dissonant chord:

4.6 Constraint Satisfaction

A computation follows a method to achieve a goal that satisfies certain constraints. In procedural
languages, the programmer states the method in detail and leaves the goal unstated, except
perhaps in the comments and documentation. In nonprocedural languages, the programmer states
the goal and lets the computer system select an appropriate method. In either kind of language,
the constraints must always be specified. The SQL database language is a typical nonprocedural
language : the programmer specifies a goal in the select clause of a query and the logical
constraints in the where clause; then the database system determines an' appropriate
method for satisfying the constraints.

By itself, a constraint is a proposition stated in some logic-based language. What makes a

constraint different from an ordinary statement is the ulterior purpose behind it some agent has
declared that the constraint must be true. Computationally, some goal triggers a search for a
combination of values that make the constraint true. The method of searching depends on both
the goal and the constraints. In SQL, for example, very different search methods would be
generated by different select options with the same where clause.

GENERATE AND TEST. For his system of rotating disks, Ramon Lull discovered a general
method for solving constraint-satisfaction problems: generate and test.
Figure 4. 1 6 shows the two-step flow chart:

Downloaded by g.sirisha Siri ([email protected])

 lOMoARcPSD|15883110

Downloaded by g.sirisha Siri ([email protected])