Dynamic ontology-based user modeling in personalized information retrieval system

Personalized retrieval is the inevitable trend of information services, which can meet users' diversified and specialized knowledge requirements. The learning of user features and the building of user model are essential to realize the personalization in information retrieval system. In this paper, we introduced ontology into this research, and study the representation of user interests using concepts and relations, and based on which, develop a dynamic approach of modeling. We analyzed the main principle and the two key steps of this approach including user features primitive learning and deep mining. The application of the model is also discussed at the end of the paper. Our study shows that dynamic ontology-based user modeling can improve the retrieval quality by better representing, discovering and utilizing user implicit interests.