Path materialization revisited: an efficient storage model for XML data

Abstract

XML is emerging as a new major standard for representing data on the world wide web. Several XML storage models have been proposed to store XML data in di#erent database management systems. The unique feature of model-mappingbased approaches is that no DTD information is required for XML data storage. In this paper, we present a new modelmapping -based storage model, called XParent. Unlike the existing work on model-mapping-based approaches that emphasized on converting XML documents to/from database schema and translation of XML queries into SQL queries, in this paper, we focus ourselves on the e#ectiveness of storage models in terms of query processing. We study the key issues that a#ect query performance, namely, storage schema design (storing XML data across multiple tables) and path materialization (storing path information in databases). We show that similar but di#erent storage models significantly a#ect query performance. A performance study is conducted using three data sets and query sets. The experimental results are presented. Keywords: Semistructured data, XML database 1