Supporting XPath axes with relational databases using a proxy index

Abstract

In recent years, a plethora of work has been done to develop methods for managing XML documents using relational databases. In order to support XPath or any other XML query language, the relational schema must allow fast retrieval of the parents, children, ancestors, or descendants of a given set of nodes. Most of the previous work has aimed at this goal using pre-/postorder encoding. Relying on this method, however, may lead to scalability problems, since the structural relationships have to be checked using nonequijoins, i.e., joins using < or > as their join condition. Thus, we discuss alternative methods, such as ancestor/descendant and ancestor/leaf indexes, and present a novel method, namely a so called proxy index. Our method allows us to replace nonequijoins with equijoins, i.e., joins using = as their join condition. The results of our comprehensive performance experiments demonstrate the effectiveness of the proxy index. © Springer-Verlag Berlin Heidelberg 2005.