Improving memory-based collaborative filtering by incorporating trust and item-popularity

摘要

Personalized recommendation is one of the most prevalent and effective ways to solve the problem of Information Overload, which has been widely applied in many fields such as e-commerce, social networking sites, and so on. Although a variety of recommendation algorithms have been proposed in the area of recommender system and achieved huge success, how to generate more accurate and intelligent recommendations to users is still a big challenge. This paper describes some limitations of current memory-based collaborative filtering recommendation methods and presents some corresponding improvements by incorporating two potential factors, that is, item-popularity and trust relationship among users, into the basic method. Experiments on two real-world datasets show that our proposed methods can provide a more accurate recommendation compared to the traditional memory-based collaborative filtering methods.