摘要
提出了一种基于信任网络的协同过滤推荐策略,在传统协同过滤策略中引入信任网络,将相似度和信任度结合在一起,提高推荐的准确率.实验证明,在数据稀疏的情况下该策略比传统的协同过滤推荐策略有更好的推荐效果.
Collaborative Filtering (CF) is one of the most prevalent recommendation approaches. It provides users with personalized services according to similarity of their preferences. However, the performance of traditional CF method is seriously limited due to the Sparsity problem. A new approach is proposed to deal with this problem. It introduces trust network in traditional CF process. Trust value is propagated through the trust network to match more neighbors for cold start users, and is combined with similarity to generate a compound weight to produce recommendations. Experiment shows that this method is more effective than traditional CF obviously.
出处
《西南师范大学学报(自然科学版)》
CAS
CSCD
北大核心
2008年第2期123-126,共4页
Journal of Southwest China Normal University(Natural Science Edition)
关键词
协同过滤
推荐系统
信任网络
信任度
collaborative filtering
recmmender system
trust network
trust value