Supporting Web Search with Near Keywords

The goal of information retrieval (IR) is to identify documents which best satisfy users' information need. The task of formulating an effective query becomes much more difficult when the target is the Web. Proposals on query refinement in IR, such as relevant feedback which needs to analyse the whole document database, cannot be applied to Web search. We propose a new method to support Web query refinement. Our methods is based on local analysis which clustering the search result. Unlike other clustering-base approaches, we take into consideration the distance between keywords, and guarantee no information loss. A Web search system is implemented for investigation of the feasibility. Candidate keywords generated by our method are provided to user to refine his/her query. We also confirmed the effectiveness by experiments on well known test collections.