Indexing XML Data with ToXin.

Abstract

Indexing schemes for semistructured data have been developed in recent years to optimize path query processing by summarizing path information. However, most of these schemes can only be applied to some query processing stages whereas others only support a limited class of queries. To overcome these limitations we developed ToXin, an indexing scheme for XML data that fully exploits the overall path structure of the database in all query processing stages. ToXin consists of two different types of structures: a path index that summarizes all paths in the database and can be used for both forward and backward navigation starting from any node, and a value index that supports predicates over values. ToXin synthesizes ideas from object-oriented path indexes and extends them to the semistructured realm of XML data. In this paper we present the ToXin architecture, describe its current implementation, and discuss comparative performance results.