Case Based Reasoning for Multi-Method Information Retrieval and Filtering

Period of Performance: 12/04/1998 - 08/15/1999

$99K

Phase 1 SBIR

Recipient Firm

Stottler Henke Associates
1650 South Amphlett Boulevard, Suite 300
San Mateo, CA 94402
Principal Investigator

Abstract

We present an innovative combination of artificial intelligence and human computer interaction techniques in the design of a new means of multimedia database access. By drawing on our experience with case base reasoning (CBR) and information retrieval and filtering, we have devised an eclectic multi-method approach that will greatly improve user efficiency and effectiveness in information search. We propose to support users with new query mechanisms that turn the major weakness of novice users' queries (i.e., imprecision) into a strength through the use of new relevance feedback and collaborative filtering techniques. Our system will also combine the power of traditional query mechanisms with new heuristic methods to enable far broader coverage than would otherwise be possible. This approach will be further augmented by allowing contextual information (e.g., current task and user profile) to influence information retrieval operations. The resulting tool will augment the user's reasoning capabilities by allowing intuitive access to information and a significantly improved feeling of control. The prototype tool developed in our Phase I effort will from the basis for the Phase II development of a complete implementation of C-MIRS (Case-based Multi-method Information Retrieval System).