融合用户标签和关系的微博用户相似性度量

Similarity Measurement of Micro-blogging Users Merging User Tags and Relationships

已有的微博用户相似性度量主要依据用户关系,实际上,在微博网络中,用户的标签信息直接表征了用户的兴趣爱好,是影响微博用户相似度的另一因素,为此,在网页相似度计算的基础上提出了融合用户关系和标签的微博用户相似性度量方法,该方法分别计算用户的链入标签相似度和链出标签相似度,并将其进行线性调和。实验从新浪微博采集实验数据,实验结果表明新方法对微博用户分类的准确率明显高于仅考虑用户关系的微博用户相似性计算方法。

Existing measurement methods of micro-blogging usersˊsimilarity are mainly based on user relationships. In fact, user tags can directly characterize userˊs interests, and is another impacting factor of user similarity. A new similarity measurement method of micro-blogging users based on webpage similarity computation is given, which effectively integrates user relationships and tags. The new method firstly computes the similarity of link-into tags and link-out tags, and then obtaines the final similarity by liner meditating the two. The experiment result based on data collected from Sina Weibo indicates that the classification accuracy of the new method is obviously higher than the method that only considers user relationships.