CONSENSUS THEORY AND EXPERT SYSTEMS

A VARIETY OF ALTERNATIVE APPROACHES HAVE BEEN PROPOSED FOR INFERENCE IN EXPERT SYSTEMS: BAYESIAN PROBABILITIES, BELIEF FUNCTIONS, FUZZY SETS, AND OTHERS. THEY DIFFER IN THE CONCEPTS THEY SEEK TO ADDRESS IN THEIR NORMATIVE JUSTIFICATION, IN EASE OF USE, AND PSYCHOLOGICAL APTNESS FOR THE PURPOSES OF EXPERTS AND EXPERT SYSTEM USERS. COMBINING RESEARCH EFFORTS IN STATISTICAL INFERENCE THEORY, COMPUTER SCIENCE, AND COGNITIVE PSYCHOLOGY, DSC PROPOSES TO ADDRESS THE PROBLEM OF DESIGNING AND SELECTING INFERENCE FRAMEWORKS FOR EXPERT SYSTEMS. THE PRIMARY OBJECTIVE OF THE INITIAL RESEARCH (PHASE I) IS TO EXPLORE THE FEASIBILITY OF DEVELOPING NEW METHODS FOR INFERENCE IN EXPERT SYSTEMS,WHICH ADDRESS SPECIFIC SHORTCOMINGS IN EXISTING APPROACHES AND COMBINE SOME OF THEIR DISTINCT VIRTUES. THE SECOND OBJECTIVE OF PHASE I RESEARCH IS TO IMPROVE UNDERSTANDING OF THE CHARACTERISTICS OF EXPERTS, USERS, AND PROBLEM DOMAINS THAT MAKE CHOICE OF ONE INFERENCE SCHEME MORE APPROPRIATE THAN ANOTHER. FIVE SUBSTANTIVE TASKS HAVE BEEN PROPOSED: (1) A SYSTEMATIC EXAMINATION OF ALTERNATIVE THEORIES OF INFERENCE IN TERMS OF FACTORS RELEVANT TO EXPERT SYSTEM APPLICATIONS; (2) THEORETICAL DEVELOPMENT OF IMPROVED INFERENCE METHODS; (3) SELECTION OF AN ARMY OPERATIONAL TESTBED; (4) COMPUTER IMPLEMENTATION OF A SMALL-SCALE EXPERT SYSTEM FOR EXPERIMENTAL TESTING; AND (5) EXPERIMENTAL COMPARISON OF ALTERNATIVE INFERENCE SCHEMES.