Using Temporal Bayesian Networks to Model user Profile Evolution

Modeling the user profile can be the first step towards personalization of information search. The user profile refers to his/her interests built across his/her interactions with the information retrieval system. It could be inferred from the recent search history limited to a single search session, during a short period of time to model short term user interests. On the other hand, from the whole search history, to model long term ones stable for a long time. In this paper, we present a personalized information retrieval approach for building and updating the user profile, based on Temporal Bayesian network. The theoretical framework provided by these networks allows better capturing and exploiting the change of user interests over time. Experiments carried out on TREC-1 ad hoc and TREC 2011 session Track collections show that our approach achieves significant improvements over a personalized search approach described in the state of the art and also to a baseline search information process that do not consider the user profile.