An effective query pruning technique for multiple regular path expressions

Abstract

Regular path expressions are essential for formulating queries over the semistructured data without specifying the exact structure. The query pruning is an important optimization technique to avoid useless traversals in evaluating regular path expressions. While the previous query pruning optimizes a single regular path expression well, it often fails to fully optimize multiple regular path expressions. Nevertheless, multiple regular path expressions are very frequently used in nontrivial queries, and so an effective optimization technique for them is required. In this paper, we present a new technique called the two-phase query pruning that consists of the preprocessing phase and the pruning phase. Our two-phase query pruning is effective in optimizing multiple regular path expressions, and is more scalable and efficient than the combination of the previous query pruning and post-processing in that it never deals with exponentially many combinations of sub-results produced from all the regular path expressions.