THE FEASIBILITY OF USING BELIEF FUNCTIONS IN DEALING WITH UNCERTAINTY IN EXPERT SYSTEMS IS EXAMINED.

THE FEASIBILITY OF USING BELIEF FUNCTIONS IN DEALING WITH UNCERTAINTY IN EXPERT SYSTEMS IS EXAMINED. THE TWO KEY AREAS FOR PROPOSED RESEARCH ARE THE PROPAGATION OF UNCERTAINTY AND THE COMBINATION OF UNCERTAIN INFORMATION. THE PROPOSED RESEARCH ON THE PROPAGATION OF UNCERTAINTY WILLBE BASED ON GENERALIZING THE NOTION OF IF-THEN FULES TO PROBABILISTIC TRANSITIONS; THE PROPOSED RESEARCH ON COMBINING UNCERTAIN INFORMATION WILL INVOLVE INVESTIGATING AN APPROACH TO DEMPSTER'S RULE WHICH AVOIDS THE PROBLEM OF NON-INDEPENDENCE IN THE EVIDENCE BEING COMBINED. THE IMPLICATIONS OF THIS RESEARCH ARE FAR REACHING ARE FAR REACHING IN SHORTENING THE DEVELOPMENT TIME OF INDIVIDUAL EXPERT SYSTEMS AND IN EXTENDING THE APPLICABILITY OF THESE SYSTEMS.