A Case-Based Reasoning Approach for Associative Query Answering

Traditional DBMS only retrieves data that perfectly match the user query and also requires the user to know the detailed database schema. Often, it is desirable to obtain additional relevant information to a query. In this paper, we present a method to provide useful information to the user that he does not explicitly asked for. Such domain specific knowledge associated to a given query depends on each user's goal and knowledge. Thus, we propose using the Case Based Reasoning paradigm to integrate past user experience and the current goal in order to guide the association. Useful associations are incrementally acquired from observations of past experiences and adapted to answer the current user query. A prototype of the associative query answering system using the proposed method has been implemented on top of the cooperative data-base system, CoBase, at UCLA. Our preliminary experimental results reveal that it is a feasible and scalable method for association control.