PRESY: A Context Based Query Reformulation Tool for Information Retrieval on the Web

Abstract

Problem statement: The huge number of information on the web as well as the growth of new inexperienced users creates new challenges for information retrieval. It has become increasingly difficult for these users to find relevant documents that satisfy their individual needs. Certainly the current search engines (such as Google, Bing and Yahoo) offer an efficient way to browse the web content. However, the result quality is highly based on uses queries which need to be more precise to find relevant documents. This task still complicated for the majority of inept users who cannot express their needs with significant words in the query. For that reason, we believe that a reformulation of the initial user's query can be a good alternative to improve the information selectivity. This study proposed a novel approach and presents a prototype system called Profile-based Reformulation System (PRESY) for information retrieval on the web. Approach: It used an incremental approach to categorize users by constructing a contextual base. The latter was composed of two types of context (static and dynamic) obtained using the users' profiles. The architecture proposed was implemented using .Net environment to perform queries reformulating tests. Results: The experiments gave at the end of this article show that the precision of the returned content is effectively improved. The tests were performed with the most popular searching engine (i.e., Google, Bind and Yahoo) selected in particular for their high selectivity. Among the given results, we found that query reformulation improve the first three results by 10.7 and 11.7% of the next seven returned elements. So as we could see the reformulation of users' initial queries improves the pertinence of returned content. Conclusion/Recommendations: Therefore, we believed that the exploitation of contextual data based on users' profiles could be a very good way to reformulate user query. This complementary mechanism would be highly improve the quality of information retrieval on the web. In the other side, we believe that more the user's profiles are properly constructed more the returned documents are relevant. Thus, the approach of constructing profiles needs to be deeply studied in order to have more representative elements. Additional data like historical searching and browsing activity of a user can be also combined to improve the query reformulation. This constitutes a part of our perspectives to improve PRESY.