CaSe - ased Retrieval Interface A itiate Dialogues in

Help desk systems have become increasingly important in the efforts of corporations to maintain customer satisfaction, and Case-Based Reasoning (CBR) p rovides promising techniques for use in the improvement of such systems. This paper describes multiple interface modes by which customer service operators can respond rapidly to customer-initiated inquiries, retrieving/storing case data from/into a case-base. The proposed interface addresses a major situation assessment problem, the difficulty of attempting to match what may be completely different descriptions of an item to be retrieved, i.e. descriptions resulting from vastly differing points of view. The proposed interface and the similarity assessment algorithm are implemented in the CARET case-based retrieval tool operating on commercial Relational Database Management Systems (RDBMS).