An Integrated Architecture for Personalized Query Expansion in Web Search

In this paper we present an integrated architecture to perform personalized interactive query expansion in Web search. Our approach is to extract expansion terms in a three stage cycle: 1) keyword extraction with lo-cal analysis on search results, 2) keyword extraction with a recommender system on a community of users and 3) an algorithm to personalize the final list of sug-gestions. Three methods for local analysis and recom-mender models are presented and evaluated. The user profile is built with latent semantic analysis on search sessions. We prepared a set of experimental scenarios with user studies in order to assess the system. The results obtained shows good performance on the per-ceived quality by users on suggested terms