Finding Hubs and Authorities using Information Scent to Improve the Information Retrieval precision.
Suruchi Chawla, Punam Bedi · International Conference on Artificial Intelligence · 2008
Improvement in the effectiveness of Information Retrieval on the Web in one such research area in which efforts are going on to improve the precision by satisfying the user need effectively and efficiently. To improve the precision of Information Retrieval on the Web there is a need to understand the information need of the query issued by user and retrieve the good authorities and hub web pages for a particular information need implied by the query. Good authorities are those pages which are rich source of information for a specific information need. Good Hub pages are those pages which contain link to good authorities for a given information need. This paper identifies the good authorities and hub pages for a particular information need using Information Scent in HITS with query sessions mining. Information scent is used to identify the relevance of the clicked URLs in the query sessions of the user with respect to the information need of the query sessions. The approach proposed in this paper uses information need of query sessions modeled using information scent and content vector of clicked URLs in the sessions to cluster query sessions with similar information need .Each derived cluster is used in HITS algorithm to generate good hubs and authorities pages for each information need represented by cluster using information scent. The input query is then used to select the cluster that contains the high scent authorities and hub page for the information need associated with the input query. Performance of the proposed approach is evaluated with an experimental study of query sessions mining of the Google Search engine web history and the experimental results shows the improvement of the Information Retrieval precision.