Page 158 - DCAP310_INTRODUCTION_TO_ARTIFICIAL_INTELLIGENCE_AND_EXPERT_SYSTEMS

Introduction to Artificial Intelligence & Expert Systems

Notes Rules of this type, often called production rules, can only produce actual behavior with the help
of an interpreter, a mechanism which applies the rules to reach a given goal. In addition to the
rule-base (which acts as a kind of long-term memory), a production rule system also has a
short-term memory (working memory) which registers the current state of the computation, as
well as current input and output states. The control structure of a production system interpreter
consists of a cyclical process, where each cycle consists of three phases:
1. Identification: This phase determines for which rules the condition sides are currently
satisfied in working memory.
2. Selection: It will often happen that more than one rule’s condition side will be satisfied.
Since in general it is not desirable for all applicable rules to fire, one or more rules are
selected on the basis of a particular conflict resolution strategy.
3. Execution: The action part of the chosen rule is executed. Although actions can take many
forms, the most typical ones involve the addition to or removal from working memory of
certain facts.

This interpreter’s mode of operation is called forward chaining or data-driven: rules are identified
when states in working memory match their conditions; the execution of the rules may in their
turn activate other rules, until a goal is achieved. But it is also possible to run an interpreter in
a backward chaining or goal-driven mode: in that case, rules are identified when their actions
match the current goals; their execution may add elements of their conditions as new goals
when they are not present in working memory, and so on, until rules are found whose conditions
match the current states in working memory. It is evident that both modes represent different
kinds of search. Rule-based architectures have been further developed toward more sophisticated
cognitive architectures, for example, ACT* (Anderson, 1983) and SOAR.

Notes The ACT* system has a semantic network (see above) as part of its long term
memory. Production systems have been used in a few psycholinguistic models, but no
models based.

Anderson, Kline and Lewis (1977) describe a production system model of language processing.
In PROZIN (Kolk, 1987), agrammatism effects are simulated by manipulating the processing
speed of the production system interpreter and the decay rate of facts in working memory.
Lewis (1993) describes a computer model of human sentence comprehension implemented in
SOAR.

Self Assessment

State whether the following statements are true or false:
3. Production systems are rule-based systems developed during the seventies as models for
human problem solving.
4. This interpreter’s mode of operation is called upward chaining.

8.3 Logic

Logic has often been used as a formal foundation for knowledge representation in AI. The
formal properties of logic formalisms are relatively well understood and make them ideally
suited as a language to which other formalisms can be translated in order to evaluate and

152 LOVELY PROFESSIONAL UNIVERSITY