摘要
好友推荐可以帮助用户发现他们感兴趣的好友,减轻信息过载的现象。然而,目前现有的推荐方法仅考虑用户链接或内容信息,其推荐精度不高,不足以提供高质量的服务。考虑了用户之间的链接和内容信息,提出了一种结合非负矩阵因式分解的主题社区好友推荐算法(T-NMF)。该算法给出了主题社区和综合相似度计算方法,产生好友推荐列表。实验表明,该算法可以更好地反映用户的偏好,并且具有比传统方法更好的推荐性能。
Friend recommendation can help users find friends which they are interested,and alleviate the phenomenon of information overload. However,the existing recommendation methods only consider user link or content information,and the recommendation accuracy is not high,so it is not enough to provide high quality service. This paper considered user link and content information,proposed a topic community recommendation algorithm based on non negative matrix factorization. The algorithm gave calculation methods of the topic community and comprehensive similarity and produced the list of friends recommended. Experiments show that the algorithm can get better recommendation performance than traditional methods.
作者
杨丰瑞
郑云俊
张昌
Yang Fengrui;Zheng Yunjun;Zhang Chang(Institute of Applied Communication Technology,Chongqing University of Posts ＆ Telecommunications,Chongqing 400065,China;Chongqing University of Posts ＆ Telecommunications Information Technology（Group）Co.Ltd.,Chongqing 401121,China)
出处
《计算机应用研究》
CSCD
北大核心
2018年第12期3624-3627,共4页
Application Research of Computers
关键词
社交网络
非负矩阵因式分解
主题社区
好友推荐
social network
nonnegative matrix factorization
topic community
friend recommendation
