Compressing XML data streams with DAG+BSBC

Abstract

Whenever the growing amount of XML data that has to be stored, processed, exchanged, or transmitted becomes a major cost driver or performance bottleneck, XML compression is an important way to reduce these problems. However, many applications, e.g. those exchanging XML data streams, also require efficient path query processing on the structure of compressed XML data streams. We present an XML compression technique called DAG+BSBC, which extends Bit-Stream-Based-Compression (BSBC) [3] by a sparse index to compressed constants that reflects DAG pointers. Furthermore, DAG+BSBC supports XML stream compression, queries on compressed data, and provides a compression ra tio that not only significantly outperforms that of other queriable XML compression tech niques, like XGrind, but is also very competitive compared to non-queriable compression tech niques like gzip and XMill. © 2009 Springer Berlin Heidelberg.