Introducing a Personal Semantics in the User Profile Modeling for a Personalized Information Retrieval

摘要

The objective of personalized information retrieval systems is to infer the real user information need hidden behind his query, and then tailor the search results to this need. The user information need may be relative to one of the user's multiple search contexts, such as user current location, user social context or an important event that occupies the public opinion of the user community. In this paper we present a new personalized information retrieval approach in which we design and use a user profile. This profile takes into account the different social, situational and temporal contexts of the user previous searches, and we introduced a personal semantics in the representation of these search contexts. The introduced personal semantics is interested by the identification and the valorization of the different cooccurrence relationships between the different relevant components of the user search history. The user profile, boasting its semantic vision, is used to infer the user current information need by the enrichment of the initial user query without going beyond the user current interest.