Testing intersection of XPath expressions under DTDs

XPath expressions play a central role in both querying XML databases and searching for XML data which is distributed over the Web. Our work focuses on XPath query optimization for XML data, which is distributed over several nodes in a database or several nodes on the Web, and which follows a single global schema defined by a DTD, i.e. multiple nodes contain fragments of a global ´virtual´ XML document. Whenever the accessible XML content stored in a node is described by an XPath expression, it may be considerably advantageous to search for XML data only on those nodes which contain a relevant fragment of the distributed ´virtual´ XML document. To enable this optimization, we present an intersection test for XPath expressions that works on the DTD and on the XPath expressions alone, i.e. it does not require any access to XML fragments. The key idea is to construct a graph containing all paths selected by one XPath expression and to show that none of these paths can be selected by the other XPath expression in any document that is valid according to a given DTD.