Problems of LCA and Impact of ORA-semantics in XML Keyword Search
No Thumbnail Available
Date
2012-03-26T02:13:57Z
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Most keyword search approaches for data-centric XML documents are based on the computation of Lowest Common Ancestors (LCA). However, LCA-based search methods depend much on hierarchical structures of XML data. Therefore it may not be able to find desired answers for many keyword queries since a relationship among objects in XML data can be represented in different hierarchical structures. In this paper, we first point out serious problems of the LCA-based approach, due to its unawareness of semantics of object, relationship and attribute, referred to as ORA-semantics. Through detailed analysis of these problems, we show the impact of ORA-semantics in XML keyword search. We then propose an ORA-semantics based approach with rules to infer expected answers for XML keyword queries. Experimental results show that our ORA-semantics based approach can resolve the problems of the LCA-based approach, and thus can be a promising research direction for XML keyword search.