Abstract:Because XML (extensible markup language) is self-described, there is much redundant structural information in XML data stream. How to compress XML data so as to reduce the network transfer cost and support XPath evaluation on the compressed data is a new area of research. The existing methods on XML compression require the multi-pass scan on data or can not support real time query processing on compressed data. In this paper, a novel compression method XSC (XML stream compression) is proposed to compress and decompress XML stream in real time. XSC constructs XML element event sequence dictionary and outputs the related index dynamically. When DTD is available, XSC can generate the XML element event sequence graph for producing more reasonable encoding before XML data stream is processed. The compressed XML data stream can be decomposed directly for XPath evaluation. Experimental results show that XSC outperforms other methods in compression ratio and compression efficiency, and the cost of XPath evaluation on compressed data stream is acceptable.