HID: An efficient path index for complex XML collections with arbitrary links

Abstract

The increasing popularity of XML has generated a lot of interest in query processing over graph-structured data. To support efficient evaluation of path expressions structured indexes have been proposed. However, most variants of structures indexes ignore inter- or intra-document references. They assume a tree-like structure of XML-documents. Extending these indexes to work with large XML graphs and to support intra-or inter-document links requires a lot of computing power for the creation process and a lot of space to store the indexes. Moreover, the efficient evaluation of ancestors-descendants queries over arbitrary graphs with long paths is a severe problem. In this paper, we propose a scalable connection index that is based on the concept of 2-hop covers as introduced by Cohen el al. The proposed algorithm for index creation scales down the original graph size substantially. As a result a directed acyclic graph with a smaller number of nodes and edges will emerge. This reduces the number of computing steps required for building the index. Thus, computing time and space will be reduced as well. The index also permits to efficiently evaluate ancestors-descendants relationships. Moreover, the proposed index has a nice property in comparison to most other work; it is optimized for descendants-or-self queries on arbitrary graphs with link relationships. © Springer-Verlag Berlin Heidelberg 2005.