Path-partitioned Encoding Optimizes Linear Twig Queries on RDBMS

Finding all the occurrences of a twig query in an XML database is a core operation for efficient evaluation of XML queries. Since every complex twig query consists of linear ones, it is important to handle linear twig queries effectively. In this paper, we propose a novel path-partitioned encoding scheme, and find a holistic solution to match linear twig queries with no structural join. We show that path-partitioned encoding scheme guarantee the I/O and CPU optimality for linear twig queries. Our experiments on representative data sets indicate that the proposed solution performs significantly.