Information Technology Reference
In-Depth Information
End-User
EXPERT SYSTEM
User-Interface
Explanation
Subsystem
Inference Engine
Knowledge
Base
Working
Memory
bi-directional information flow
Fig. 2.14
An expert system architecture
• Workingmemory:Adataareausedforstoringtheintermediateorpartialresults
of problem solving.
• Userinterface:Aninterfacethatallowsend-userstointeract with the ES.
• Explanationsubsystem:Asetoffacilitiesthatenabletheusertoaskquestionsof
the system, about how, for instance, the system came to a particular conclusion.
2.6.1
Knowledge Representation
A general model for knowledge representation is to form the basis of a system
exhibiting human intelligence. Such a model is likely to require a wide variety of
knowledge representation formalisms to represent different types of knowledge
such as current facts, past and future knowledge, meaning of words, certain and un-
certain situations, negative situations, etc. There are several schemes for represent-
ing knowledge in an ES. The most common methods of knowledge representation
are semantic networks, rule-based systems, and frame-based systems.
1. Semantic Networks
The most general representational scheme, and also one of the oldest in AI, is the
semantic network (or semantic net). A semantic network is an explicit taxonomic
Search WWH ::
Custom Search