| Summary: | 碩士 === 國立彰化師範大學 === 資訊管理學系所 === 101 === Recommender system has been a hot topic and attracted much attention in both research and practice in recent years. Two broad classes of recommendation techniques that are commonly used in current recommender systems are content-based filtering and collaborative filtering. Both approaches utilize user’s preference for making personal recommendation. In addition to user’s preference, however, in reality, a user’s decision to buy a product is often influenced by her acquaintances. Therefore, in this research, we combine user’s preference and social relationship to make personal recommendation. Four variations of the proposed approach were tested using rating and social data downloaded from the epinions system (http://www.epinions.com). The results from our experimental evaluations demonstrate that our proposed IS approach outperforms the SS, SI, CR, and II approaches.
|
|---|
