PASSIVE KNOWLEDGE ACQUISITION

Period of Performance: 01/01/1991 - 12/31/1991

$500K

Phase 2 SBIR

Recipient Firm

Software Productivity
122 4th Ave.
Indialantic, FL 32903
Principal Investigator
Firm POC

Abstract

THE DEVELOPMENT OF KNOWLEDGE-BASED SYSTEMS CURRENTLY REQUIRES THE USE OF A KNOWLEDGE ENGINEER TO TRANSLATE AND ENCODE THE KNOWLEDGE EXTRACTED FROM AN EXPERT INTO THE TARGET REPRESENTATION AND REASONING SYSTEM. ALTHOUGH THE GOAL OF PROVIDING SYSTEMS WITH WHICH THE EXPERT CAN DIRECTLY INTERACT HAS BEEN SOUGHT FOR SOME TIME, THESE SYSTEMS STILL FALL WELL SHORT OF THEIR INTENDED GOAL. A MAJOR REASON IS THAT THE EXPERT MUST SHIFT THE FOCUS OF THEIR COGNITIVE PROCESSING BETWEEN THE PERFORMANCE OF THE TASK TO BE CAPTURED AND THINKING ABOUT HOW TO ENCODE THAT PROCESS. THIS EFFORT WILL DEVELOP A PASSIVE KNOWLEDGE ACQUISITION SYSTEM WHICH UNOBTRUSIVELY MONITORS AND RECORDS THE EXPERT'S ACTIONS. THEN, USING A COLLECTION OF ELICITATION AND DISCOVERY HEURISTICS, THE SYSTEM QUERYS THE EXPERT REGARDING THE RATIONALE AND DECISION PROCESS BEHIND THE ACTIONS. THIS APPROACH WILL PROVIDE AN EXPERT KNOWLEDGE ELICITATION SYSTEM WHICH BEHAVES MUCH THE SAME AS A HUMAN KNOWLEDGE ENGINEER.