Detecting session boundaries to personalize search using a conceptual user context

Abstract

Most popular Web search engines are carachterized by "one size fits all" approaches. Involved retrieval models are based on the query-document matching without considering the user context, interests ang goals during the search. Personalized Web search tackles this problem by considering the user interests in the search process. In this chapter, we present a personalized search approach which adresses two key challenges. The first one is to model a conceptual user context across related queries using a session boundary detection. The second one is to personalize the search results using the user context. Our experimental evaluation was carried out using the TREC collection and shows that our approach is effective. © 2009 Springer Netherlands.