Value-based predicate filtering of XML documents

Abstract

In recent years, publish–subscribe systems based on XML filtering have received much attention in ubiquitous computing environments and Internet applications. The main challenge is to process a large number of content against millions of user subscriptions. Several XML filtering systems focus on the efficient processing of structural matching of user subscriptions represented as XPath twig patterns. However, existing techniques provide limited or no support for twig patterns that contain various operators in the value-based predicates. In this paper, we present the pFiST system that filters XML documents by transforming twig patterns into sequences based on Prüfer’s method. This sequencing idea for XML filtering was first demonstrated by FiST [J. Kwon, P. Rao, B. Moon, S. Lee, FiST: scalable XML document filtering by sequencing twig patterns, in: Proceedings of the 31st VLDB Conference, Trondheim, Norway, 2005, pp. 217–228]. The focus of pFiST is to support value-based predicates in twig patterns in addition to matching their structure. The pFiST system supports equality and non-equality operators, and in addition can handle logical operators such as AND and OR in the value-based predicates. Extensive experimental results show that pFiST provides good performance over data sets with different characteristics.