Keyword search is an effective paradigm for information discovery and has been introduced recently to query extensible markup language (XML) documents. Due to the existence of ambiguities in query semantics, effective keyword search of XML documents needs to infer search intention. In this study, a semantics oriented approach for keyword search intention inference over XML documents is proposed. The method infers search intention in two steps: in the first step, XML nodes are mapped into the concepts in ER model and the content features, occurrence features and reference features are utilized to classify the nodes into different categories. In the second step, query-dependent features are incorporated to refine search intentions. An efficient algorithm is also proposed to process keyword queries. Experimental results demonstrate the effectiveness of the proposed methods.
展开▼