A Collaborative Filtering Method Based on Both the Nearest and Potential Neighbors
In the traditional collaborative filtering algorithm, the predictions for the unrated items are generated based on only the ratings given by the nearest neighbors of the target user. The problem of the traditional method is that the rating sparsity may lead to an unreliable recommendation. In this paper, we propose a new method to overcome this problem by taking the potential neighbors of the target user into consideration. Using the transfer similarity, we find out the potential neighbors of the target user and generate the predictions based on both the nearest and the potential neighbors. The experiment results demonstrate that our method can help to improve the accuracy and the coverage of the collaborative filtering algorithm.
collaborative filtering nearest neighbors potential neighbors
Jing Miao Guicai Wang Zhaoguo Xuan
Institute of Systems Engineering, Dalian University of Technology, Dalian, China
国际会议
广州
英文
74-78
2008-12-11（万方平台首次上网日期，不代表论文的发表时间）