Specific Academic Area based Automatic Query Expansion

Most of the current search engines focus on the content-based searching; therefore, they require the users to input accurate query terms. Unfortunately, most people would like to input just a few words when doing the search and the words themselves might have diverse meanings, which badly decrease the searching performance. The previous automatic query expansion algorithms usually use the documents from all of the fields to calculate the similarity between terms. At the same time, they didn´t consider the non-correlation between terms to expand the query. Because the interests of users are often limited to a relative narrow branch, the global relationships among terms usually introduce some irrelevant noise. Therefore, if users can choose some relevant area and then do the search, the results can largely fulfill the users´ requirements. In this paper, we choose a specific academic area and add the non-correlation into query expansion. Our experiments have well demonstrated the effectiveness of category-based search.