Rule Based Database Integration in HIPED : HeterogeneousIntelligent Processing in Engineering

In this paper 1 we describe one aspect of our research in the project called HIPED, which addressed the problem of performing design of engineering devices by accessing heterogeneous databases. The front end of the HIPED system consisted of interactive KRI-TIK, a multimodal reasoning system that combined case based and model based reasoning to solve a design problem. This paper focuses on the backend processing where ve types of queries received from the front end are evaluated by mapping them appropriately using the \facts" about the schemas of the underlying databases and \rules" that establish the correspondance among the data in these databases in terms of relationships such as equivalence, overlap and set containment. The uniqueness of our approach stems from the fact that the mapping process is very forgiving in that the query received from the front end is evaluated with respect to a large number of possibilities. These possibilities are encoded in the form of rules that consider various ways in which the tokens in the given query may match relation names, attrribute names, or values in the underlying tables. The approach has been implemented using CORAL deductive database system as the rule processing engine.