Abstract

As the popularity of areas including document storage and distributed systems continues to grow, the demand for high performance XML databases is increasingly evident. This has led to a number of research eorts aimed at exploiting the maturity of relational database systems in order to in- crease XML query performance. In our approach, we use an index structure based on a metamodel for XML databases combined with relational database technology to facilitate fast access to XML document elements. The query process involves transforming XPath expressions to SQL which can be executed over our optimised query engine. As there are many dierent types of XPath queries, varying processing logic may be applied to boost performance not only to indi- vidual XPath axes, but across multiple axes simultaneously. This paper describes a pattern based approach to XPath query processing, which permits the execution of a group of XPath location steps in parallel.