Complex Twig Pattern Query Processing over XML Streams

Abstract

The problem of processing streaming XML data is gaining widespread attention from the research community. In this paper, a novel approach for processing complex Twig Pattern with OR-predicates and AND-predicates over XML documents stream is presented. For the improvement of the processing performance of Twig Patterns, all the Twig Patterns are combined into a single prefix query tree that represents such queries by sharing their common prefixes. Its OR-predicates and AND-predicates of a node are represented as a separate abstract syntax tree associated with the node. Consequently, all the Twig Patterns are evaluated in a single, document-order pass over the input document stream for avoiding the interim results produced by the post-processing nested paths of YFilter. Compared with the existing approach, experimental results show that it can significantly improve the performance for matching complex Twig Patterns over XML document stream, especially for large size XML documents. Based on the prior works, the optimization of twig patters under DTD (document type definition) by using structural and constraint information of DTD is also addressed, which is static, namely, it is processed before the runtime of stream processing.