Finding Missing Answers due to Object Duplication in XML Keyword Search

XML documents often have duplicated objects, with a view to maintaining tree structure. Once object duplication occurs, two nodes may have the same object as the child. However, this child object is not discovered by the typical LCA (Lowest Common Ancestor) based approaches in XML keyword search. This may lead to the problem of missing answers in those approaches. To solve this problem, we propose a new approach, in which we model an XML document as a so-called XML IDREF graph so that all instances of the same object are linked. Thereby, the missing answers can be found by following these links. Moreover, to improve the efficiency of the search over XML IDREF graph, we exploit the hierarchical structure of the XML IDREF graph so that we can generalize the efficient techniques of the LCA-based approaches for searching over XML IDREF graph. The experimental results show that our approach outperforms the existing approaches in term of both effectiveness and efficiency.