Improving View Selection in Query Rewriting Using Domain Semantics

Query rewriting using views is an important issue in data integration. Several algorithms have been proposed, such as the bucket algorithm, the inverse rules algorithm, the SVB algorithm, and the MiniCon algorithm. These algorithms can be divided into two categories. The algorithms of the first category are based on use of buckets while the ones of the second category are based on use of inverse rules. The bucketbased algorithms have not considered the effects of integrity constraints, such as domain semantics, functional and inclusion dependencies. As a result, they might miss query rewritings or generate redundant query rewritings in the presence of these constraints. A bucket-based algorithm consists of two steps. The first step is called view selection that selects views relevant to a given query and puts the views into the corresponding buckets. The second step is to generate all the possible query rewritings by combining a view from each bucket. In this paper, we consider an improvement of view selection in the bucket-based algorithms using domain semantics. We use the resolution method to generate a pseudo residue for each view given a set of domain semantics. Given a query, the pseudo residue of each view is compared with it and any conflict that exists can be found. As a result, irrelevant views can be removed even before a bucketbased algorithm is used.