摘要
社会网络中包含大量的社会信息,如何从这些社会信息中发掘对用户有用的信息已成为学者和专家的研究热点。本文提出一种基于社会正则化的推荐算法:把改进的矩阵分解技术应用到社会化推荐中;利用社会网络中用户间的朋友关系来优化对用户的建模,学习更好的用户特征空间模型;利用社会网络中的标签信息建立用户和物品的关系,并利用这种关系来优化用户-物品的建模。实验结果表明,改进后的推荐算法的精确度高于传统的推荐算法,有效地解决了社会信息冗余问题。
Social network includes vast amount of social information, how to find information users are interested in has become research focus of many scholars and experts. Based on this idea, this paper proposes a social regularization-based recommendation algorithm： apply the matrix factorization technology into the social recommendation, make use of friendship between users to get better user＇ s feature space, consider the tag information in social network, and use this information to learn better user and item feature space. The analysis of experiments shows that the accuracy of the improved algorithm is better than the traditional recom- mendation algorithm and it solves the problem of redundant social information effectively.
出处
《计算机与现代化》
2014年第1期77-80,共4页
Computer and Modernization
关键词
矩阵分解技术
社会化推荐
特征空间模型
社会信息
matrix factorization technology
social recommendation
feature space model
social information
