Algorithms for Answering Queries Eeciently Using Views

Algorithms for answering queries using views have been used to integrate information from multiple sources. The bucket algorithm, predominantly used to reformulate queries, has two drawbacks. It generates a large number of candidate solutions, even when the number of sound solutions is small, and then checks for their soundness using an exponential conjunctive-query-containment test. Leser 1] presents an algorithm that generates a smaller number of candidate solutions and also eliminates the conjunctive-query-containment-test, but does not guarantee the soundness of all generated solutions. We introduce the concept of shared variable buckets(SVB) covering all subgoals that share a variable in order to ensure that only sound solutions are generated and thus avoid the conjunctive-query containment test. The algorithm presented has the advantage that it avoids generation of a large set of candidate solutions whenever possible. Experimental results prove the superiority of this algorithm over previous solutions.