抄録
This paper proposes an advanced movie recommender system. The system is primarily based on the content-based collaborative filtering or hybrid filtering technique. The distinctive point of this system lies in an improved neighborhood formation method in order to get high-quality neighbors. Current hybrid systems only use user's opinions as user's rating values in selecting neighbors. That causes loss of user preference features, and tends to provide poor quality neighbors. The proposed system uses the user's opinions on various features of user preferences in selecting neighbors. That results in high-quality neighbors and high-quality recommendations will be obtained accordingly. An experimental movie recommender system, called Advanced Yawara system has been developed to prove the effectiveness of the method. The evaluation results show that the Advanced Yawara system provides higher quality of recommendations than current hybrid systems.
