一种新的基于用户群体关系挖掘的随机漫游社会推荐模型

传统的协同过滤推荐算法在用户评分稀疏时,存在冷启动问题等不足,而最近几年提出的基于信任度的推荐算法以及一些它们的混合算法虽然解决了冷启动问题,却忽略了用户群体特征.针对上述情况,利用社会网络分析方法对社会性网络中的用户群体关系进行挖掘,提出一种全新的社会推荐模型Cliqueswalk,同时给出了权威推荐,为用户提供权威(意见领袖)的参考意见.实验表明,新的算法能够大大缩小目标评分信息的查找范围,推荐效率明显优于已有的协同过滤推荐算法、基于信任度的推荐算法以及它们的混合算法.

New Social Recommendation Model of Random Walks Based on Users Groups Relation Mining

The traditional Collaborative filtering algorithms exist "Cold Start" problem when users′ data of ratings are sparse.Trust-based method and some combined methods proposed in recently years can better deal with cold start users,but all these methods ignore the characteristics of user groups in social network.Aimed at the above instances,this paper presents a new social recommendation model-Cliqueswalk.Furthermore,we also give the authoritative recommendation which provides the opinion of authoritative user(opinion leader) for users.Experiment results demonstrate that the new method can greatly narrowed the scope of searching target rating information and the effect of recommendation is better than the existing methods of Collaborative filtering,trust-based and their combined.