A Flexible Labeling Scheme for Efficient Structural Join over XML Graph

Abstract

The most important operation of querying XML documents is finding occurrences of structural relationships between tagged elements. Performing such structural joins needs to encode nodes in XML documents. Once each node is labeled by the interval-based labeling scheme, structural relationships between nodes can be determined easily in constant time. However this paper shows that traditional approaches work only for tree-structured data and cannot be applied to XML documents being modeled as arbitrary graphs. The main contribution of this paper is the solution of matching structural relationships in graph-structured XML data, under a new interval-based labeling scheme. Based on the idea of e-nodes, each XML node can be labeled (with an optional E-List) appropriately by our G_encoding algorithm for solving the multiple-inheritance and cyclic-path problems. We have formally analyzed the properties of the new labeling scheme, and developed an efficient structural join algorithm SJG using the scheme.