Research On Inferring Users’ Query Intention For XML Keyword Search

Keywords retrieval technology based on XML data has always been a hotissue researchers focus on. How to accurately infer the users’ search intention andthen identify the XML node types that user wants to search for is one of the keyproblems of XML keyword search. For a given user’s keyword query, the paperstudies how to accurately infer the user’s search intention, and then effectivelyreturn XML keyword search results. The main research contents are as follows.First, through in-depth analysis and comparison on existing methods to infersearch intention of XML keyword query, we find that existing methods have twoproblems:(1) The search intention inferred maybe correspond to many nodes whichonly contain part of the query keywords, so that the sub-trees rooted by them cannot be treated as the query results;(2) The search intention inferred maybecorrespond to many nodes, although they contain all of the query keywords, theymay contain too much irrelevant information.Secondly, to address the above problems, we propose a new method ofinferring users’ search intention by counting the results’number of various types inthe query results. It avoids the inaccurate problem of existing methods which onlyconsider the inclusion relationship between the results type and a single querykeyword.Thirdly, according to the truth that query results in the result type contain bothcontent and structure information, we propose a new ranking strategy to estimate therelevance between different results and a given query so that return high qualityquery results to improve the effectiveness of XML keyword query.Finally, a large number of experimental results verify the effectiveness of themethod we proposed.