Semantic Interoperation: A Query Mediation Approach

Query mapping and transformation techniques for problem solving with multiple knowledge servers. of data models as canonical models for federated databases. 15] W. Kent. Solving domain mismatch and schema mismatch problems with an object-oriented database programming language. 27 Much research remains to be done. First, we have focused on three components of the mediation architecture, namely the mediation language, the query mediator, and the connict detectors. Research is needed in the other components. Although it is often straightforward to translate participating schemas into the mediation language, it is tricky to translate the queries of participating databases into the mediation language if they could involve higher-order constructs, such as in the case of object-oriented databases. It is even more challenging to bind queries in the mediation language to the representational constructs of participating databases such that they could be eeciently evaluated. Second, we have assumed that the knowledge in the mediator's knowledge base is available. How to obtain such knowledge is certainly an important issue. Although the acquisition of such knowledge is likely to be a highly interactive process, automated acquisition tools would be valuable. Third, we have restricted ourselves to constraints, relationships, and queries that do not involve negation. The semantics of query mediation could certainly be generalized to allow negation, as long as for example the result is stratiied 31]. The approach could also be easily generalized to deductive databases containing rules in addition to constraints. Finally, research is needed in the autonomous optimization of mediated query evaluation. Due to the autonomy of participating databases, the query mediator often does not have access to the performance information that is crucial in query optimization. The query mediator needs a cost model that is independent of the implementation structures of participating databases. Techniques are also needed for the mediator to obtain performance information by querying 9]. References 1] Y. Arens and C. Knoblock. Planning and reformulating queries for semantically-modeled multidatabase systems.