A Novel Technique for Inferring User Search using Feedback Sessions

Web search engines have become a resource pool to gain information. We submit queries to search engines to retrieve search results. These queries basically depict information needs of users. But there are situations in which these queries do not represent user's specific information needs as varied users have different needs with respect to same query. There are many search engines like Google, Yahoo, Amazon that retrieve search results for users based on their ranking algorithm. It should also be noted that while inferring user search goals, fulfillment of user's desire is very important. So a Novel Technique to Infer User Search Goal using Feedback Session is proposed. The basic step of Inferring user search goals is generation of feedback session through user click-through logs. Feedback session distinguishes between what user requires and what the user does not require. Keywords in the query will determine whether a document satisfies the user's needs. The initial search results will later be restructured based on users' needs and manually Average Precision would be used to examine the performance of users. The main objective of this technique is to improve information retrieval by generating feedback sessions and rank the results list based on user's domain of interest. These re ranked results will be categorization based which will be dependent on the keywords generated.