Improving the Efficiency of XPath Execution on Relational Systems

This work describes a method for processing XPath on a relational back-end that significantly limits the number of SQL joins required, takes advantage of the strengths of modern SQL query processors, exploits XML schema information and has low implementation complexity. The method is based on the splitting of XPath expressions into Primary Path Fragments (PPFs) and their subsequent combination using an efficient structural join method, and is applicable to all XPath axes. A detailed description of the method is followed by an experimental study that shows our technique yields significant efficiency improvements over other XPath processing techniques and systems.