Optimizing Queries in Extended Relational Databases

Abstract

We investigate the optimization of extended relational queries used in systems holding, for example, spatial, multimedia or constraint data. For such queries we must account for the built-in relations specific to the kind of data, and application dependent relationships between different relations. We show that the constraint database perspective and the use of constrained tuple-generating dependencies provides a general framework in which to address semantic query optimization for these queries. We establish some sufficient conditions for query transformations involving the introduction of relations, extending work in the literature for conventional databases. We introduce semantic query partition (SQP) as a useful technique for optimizing queries with expensive operations, and investigate the problem of generating subqueries, which is central to the use of SQP.