Efficient evaluation of distance predicates in Xpath full-text query

Abstract

In recent years, more and more XML repositories are emerging, e.g., XML digital library, SIGMOD and DBLP document collections. Since XML is good at representing both structured and unstructured data, to facilitate the usage of this kind of information, it is necessary to support structure-based and content-based (full-text) queries/retrievals over XML repositories. With existing XPath/XQuery Pull-Text, user could do search based on cardinality, proximity or distance predicates. In this paper, we propose an efficient approach for the Information Retrieval (IR) style search, especially distance predicates search, on XML documents. Numbering technique is employed to encode XML documents, and then three algorithms are designed to evaluate queries with distance predicates. To improve the performance, some optimization techniques are introduced. Extensive experiments show the effectiveness and efficiency of the proposed approach. © Springer-Verlag Berlin Heidelberg 2006.