naistar (NAIST Academic Repository) >
学術リポジトリ naistar / NAIST Academic Repository naistar >
テクニカルレポート / Technical Report >
Please use this identifier to cite or link to this item:
|Title: ||Virtual joins for XML data|
|Authors: ||Kha, Dao Dinh|
|Issue Date: ||Nov-2003|
|Publisher: ||Nara Institute of Science and Technology|
|Series/Report no.: ||Information Science Technical Report ~ TR2003012|
|Abstract: ||Establishing the hierarchical order among the XML elements is an essential function of the XML query processing techniques and there are a number of proposals for the task. Although most of XML documents have associated DTD or XML schema, the prior query processing techniques have not utilized the document structure information efficiently. Each of prior techniques has an advantage in processing only a type of queries and it is difficult to incorporate them to complement each other. In this study, we propose a novel XML query processing method that uses DTD or XML schema to improve the I/O complexity of XML query processing. We design a Structure-based Coding for XML data (SCX) that incorporates both structure and tag name information extracted from the document structure descriptions. Given the tag name and the structure code of an element, SCX allows to determine the tag name and the structure code of the parent element without I/O. This property of SCX provides a Virtual Join mechanism that greatly reduces I/O workload for processing XML queries. We present the algorithms to apply the Virtual Joins for processing the queries of both path and twig patterns, where SCX can be integrated with other structural join techniques to improve the XML query processing. Our experimental results indicate that SCX accelerates the processing of path queries significantly and the efficiency increases in correspondence with the join workload of XML queries and the size of data sets.|
|Text Version: ||author|
|Appears in Collections:||テクニカルレポート / Technical Report|
Files in This Item:
There are no files associated with this item.
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.