Improving the Performance of Personalized Recommendation with Ontological User Interest Model

Abstract

Personalized recommendation is effective to provide good recommendations to different users to meet different needs. However, it remains a challenge to make personalized recommendation sensitive to the semantic information of a user's specific context and to the changing of user interests over time. A user interest model based on user interest ontology is proposed in this paper. The incrementally updating algorithm of user interest model is described based on Spreading Activation Theory. Using the ontological user interest model, the recommendation process is presented in detail. Using movie rating data from Movie lens, we demonstrate that this recommendation algorithm offers improved personalized recommendation performance, including measures of MEA, diversity and cold-start performance. Finally, the stability of user interest model is analyzed.