Mining users´ interest graph in social networks with topic based tag propagation

With the popularity of micro-blogging, more and more people get them involved in this social network platform. The tremendous user generated content can be used to mining user´s interest graph. As the duplicate of reality, users display their hobby through social network behaviors. In this paper, we focus on Sina Weibo, which is the most notable micro-blogging service in China. Our methodology relies on the Sina Weibo user self-tags. Usually the tags are the conclusion of him in some aspects and contain a lot of information about his interest points. However, only 20% users tag themselves. And most users have less than 3 tags. In order to mining users´ interest graph sufficiently, we propose a topic based tag propagation model. The experiment result highlights that the model gets a good performance.