MULTIPLE FORMS OF KNOWLEDGE REPRESENTATION WITHIN A SINGLE EXPERT SYSTEM

THERE ARE SEVEN PRINCIPAL METHODS OF REPRESENTING KNOWLEDGE IN AN EXPERT OR OTHERWISE INTELLIGENT COMPUTER SYSTEM, EACH METHOD BEST SUITED TO A DIFFERENT KIND OF KNOWLEDGE. THEY ARE: LOGIC, PROCEDURAL REPRESENTATIONS, SEMANTIC NETWORKS, PRODUCTION SYSTEMS, DIRECT (ANALOGICAL) REPRESENTATION, SEMANTIC PRIMITIVES, AND FRAMES AND SCRIPTS. THESE HAVE MAINLY BEEN USED INDIVIDUALLY. HOWEVER, PROBLEM-SOLVING (A MAJOR GOAL OF INTELLIGENT SYSTEMS) OFTEN REQUIRES THE USE OF MORE THAN ONE KNOWLEDGE. IS IT POSSIBLE FOR AN INTELLIGENT SYSTEM TO EMPLOY MORE THAN ONE KNOWLEDGE REPRESENTATION SCHEME AND TO USE KNOWLEDGE FROM MORE THAN ONE SCHEME IN ITS REASONING OR PROBLEM-SOLVING? PHASE I PROPOSAL IS TO SELECT TWO REPRESENTATION SCHEMES AND TO DESIGN RESEARCH PROCEDURES TO USE THEM COOPERATIVELY IN PROBLEM-SOLVING. EVAULATIONS WILL BE EXPLORED BOTH OF REPRESENTING DIFFERENT KNOWLEDGE IN DIFFERENT KNOWLEDGE REPRESENTATION SCHEMES AND OF REPRESENTING THE SAME KNOWLEDGE IN TWO DIFFERENT KNOWLEDGE REPRESENTATION SCHEMES.