FOR KNOWLEDGE-BASED SYSTEMS

USING A TWO-PROCESS KNOWLEDGE REPRESENTATION APPROACH, WE PROPOSE TO CONSTRUCT A DEFINITION OF MENTAL MODELS FOR KNOWLEDGE-BASED SYSTEMS. THIS EXPLOARTORY DEVELOPMENT PROJECT WILL ALSO DELINEATE THE SYSTEMS AND PROBLEM DOMAINS WHICH ARE BEST SUITED TO REPRESENTATION IN THIS WAY. THIS APPROACH WILL GIVE KNOWLEDGE REPRESENTATION IN EXPERT SYSTEMS A CLOSER RESEMBLANCE TO THE MENTAL MODELS THAT CHARACTERIZE HUMAN COGNITION, WHILE ALSO COMPENSATING FOR THE LIMITATIONS OF CERTAIN COGNITIVE FUNCTIONS. WE INTEND TO SHOW THAT SUCH A MODIFIED REPRESENTATION IMPROVES THE PERFORMANCE EFFICIENCY OF KNOWLEDGE-BASED SYSTEMS AND PERMITS THEM TO BE APPLIED TO LARGER AND MORE COMPLEX CON-TEXT-DEPENDENT PROBLEM DOMAINS. IN PHASE I, THE TWO-PROCESS KNOWLEDGEREPRESENTATION TECHNIQUE WILL BE DEVELOPED AND ITS FEASIBILITY WILL BEDEMONSTRATED. IMPROVEMENT OVER TRADITIONAL KNOWLEDGE REPRESENTATION SCHEMES WILL BE TESTED. IN PHASE II WE PROPOSE TO REFINE DEVELOPMENT METHODS AND TO DEVELOP A COMPLETE EXPERT SYSTEM PROTOTYPE USING THE NEW TECHNIQUE.