Accelerating XPath Evaluation against XML Streams

Data streams are an emerging technology for data dissemination in cases where the data throughput or size make it unfeasible to rely on the conventional approach based on storing the data before processing it. Areas where data streams are applied include monitoring of scientific data (astronomy, meteorology), control data (traffic, logistics, networks), and financial data (bank transactions). Querying XML data streams without storing and without decreasing considerably the data throughput is especially challenging because XML streams convey tree structured data with (possibly) unbounded size and depth. This paper presents a non-trivial, yet natural and simple extension of the SPEX query processor for XML streams [1]. This extension, called structural filters, exploits the structural relationships between the XML fragments encountered along the stream at various processing states in order to skip large stream fragments irrelevant to the query answer. Experimental evidence shows clearly the efficiency of SPEX with structural filters over the normal SPEX. These structural filters are demonstrated together with SPEX using SPEX Viewer [1], a sophisticated visualization of the SPEX run-time system. More precisely, SPEX Viewer visualizes step-by-step query rewriting, query plan generation, incremental stream processing, and answer generation.