融合链接拓扑结构和用户兴趣的朋友推荐方法

发布时间：2018-04-24 16:49

本文选题：社交网络 + 朋友关系　；参考：《通信学报》2015年02期

【摘要】：提出一种新的朋友推荐方法,该方法同时使用用户兴趣和朋友关系这2种因素来为目标用户推荐朋友,对PageRank算法进行改进,提出一种能同时融合上述2种因素的Topic_Friend_PageRank(TFPR)模型。首先,采用LDA(latent Dirichlet allocation)分析用户发布的消息内容,将用户表示为若干主题上的分布,从而建模用户的兴趣。接下来,使用加权的PageRank算法建模用户在整个链接拓扑中的重要程度和用户之间朋友关系的相似性。最后根据主题感知的PageRank思想,将用户兴趣融入前面提到的加权PageRank中,形成同时融合用户兴趣和朋友关系的TFPR模型。采用新浪微博数据验证所提模型的性能,实验证明该模型能同时得到较高的准确率和召回率。
[Abstract]:In this paper, a new friend recommendation method is proposed. The two factors of user interest and friend relationship are used to recommend friends for the target users. The PageRank algorithm is improved, and a TopicFriendPageRankP TFPRs model is proposed, which can fuse the above two factors at the same time. Firstly, the LDA(latent Dirichlet location is used to analyze the message content published by the user, and the user is represented as the distribution on several topics, so as to model the interest of the user. Then, the weighted PageRank algorithm is used to model the importance of the user in the whole link topology and the similarity of the friend relationship between the users. Finally, according to the PageRank idea of topic awareness, the user interest is integrated into the weighted PageRank mentioned above, and a TFPR model combining user interest and friend relationship is formed. The performance of the proposed model is verified by using Sina Weibo data. The experimental results show that the proposed model can obtain high accuracy and recall simultaneously.
【作者单位】：中国科学院计算技术研究所;中国科学院信息工程研究所;
【基金】：国家高技术研究发展计划(“863”计划)基金资助项目(2011AA010705) 先导专项基金资助项目(XDA 06030200) 国家自然科学基金资助项目(61003167)~~
【分类号】：TP393.092;TP391.3

【共引文献】