摘要
推荐系统已经得到了广泛的研究和应用,但是大多数推荐系统中仍存在一些导致系统推荐质量低下的不足:用户-信息项矩阵的大规模性和数据稀疏性,假设所有的用户都是互相独立的,该假设忽略了用户之间的联系。为了提高推荐系统模型的准确性,提出一种新型的概率因子分析方法。该方法对社交网络图进行挖掘,并将挖掘出的信任关系应用到推荐系统中,从而把用户朋友的喜好与用户的兴趣融合在一起,用于提高推荐质量。理论分析和实验结果表明,该方法复杂度是线性的,相对于传统方法表现出了很大的优越性,适合应用于大规模数据处理。
Recommendation system has been widely studied and applied, but most of the recommendation systems still have shortcomings ： enormous scale and sparsity of user-information item matrix, assuming all the users are independent but overlooking the links between them, these lead to the degradation of system＇s recommendation quality. In order to improve the accuracy of recommendation system model, we pres- ent a novel probabilistic factor analysis method. The method mines the social network graph, and applies the mined trust relationship to rec- ommendation system, thus integrates the preferences of user＇s friends with the interests of users for improving the recommendation quality. Theoretical analysis and experimental results indicate that the method is linear complexity, and demonstrates a significant superiority in com- parison to traditional methods, and is suitable for large-scale data processing.
出处
《计算机应用与软件》
CSCD
北大核心
2014年第7期31-35,共5页
Computer Applications and Software
关键词
推荐系统
社交网络
信任关系
矩阵分解
Recommendation system
Social network
Trust relationship
Matrix factorisation
