社交网络用户标签预测研究

发布时间：2018-09-08 20:43

【摘要】：随着社交网站的流行以及用户的大规模增加,社交网络用户行为分析已经成为社交网站进行网站维护、性能优化和系统升级的重要基础,也是网络知识挖掘和信息检索的重要研究领域。为了更好地理解社交网络用户添加个人标签的行为特征,该文基于大约263万个微博用户的真实数据,对用户标签的分布进行了研究和分析。我们主要考察了用户标签的宏观分布特征,以及用户标签与关注对象的标签分布之间的联系,发现微博用户给自己添加标签时,在开始阶段倾向于使用反映个性的标签,之后会出于从众心理而选用大众化标签。我们将研究发现运用到基于关注关系的标签预测算法中,结果证实相关分析对于社交网站的标签推荐等课题具有一定的参考意义。
[Abstract]:With the popularity of social networking sites and the large-scale increase of users, social network user behavior analysis has become an important foundation for social networking sites to carry out website maintenance, performance optimization and system upgrading. It is also an important research field of network knowledge mining and information retrieval. In order to better understand the behavior characteristics of social network users adding personal tags, this paper studies and analyzes the distribution of user tags based on the real data of about 2.63 million Weibo users. We mainly investigate the macro distribution characteristics of user tags and the relationship between user tags and tags distribution of objects of concern. We find that Weibo users tend to use tags reflecting their personality in the initial stage when they add tags to themselves. The popular label will then be chosen out of conformity with the crowd. We apply the research findings to the tag prediction algorithm based on the relationship of concern. The results show that the correlation analysis has a certain reference significance for the label recommendation of social networking sites.
【作者单位】：清华大学智能技术与系统国家重点实验室清华信息科学与技术国家实验室(筹)清华大学计算机系;
【分类号】：TP391.1;TP393.092

【相似文献】