Implicit Community Discovery Based on Microblog Theme Homogeneity

Abstract

With the rapid development of various social platforms, social networks have accumulated more and more data. These huge and complex data contain valuable information in many fields. It is of far-reaching significance to conduct statistical analysis of these data and dig out valuable information and apply it in real life. Based on the homogeneity of social networks, our paper proposes a framework of discovery algorithms based on the homogeneity of microblogs to realize the theme community discovery of microblog social networks. Firstly, the data set is preprocessed, the invalid data is removed and classified by microblog users. Secondly, the LDA model is used to extract the topic of microblog user blogs to describe the characteristics of microblog users. Thirdly, two users are calculated based on the theme interest. The homogeneity measure between them is used to represent the relationship of social networks. Finally, the community of the implicit social network is realized through unsupervised algorithms to build communities with the same interest as the theme. In order to verify the validity of our proposed algorithm framework, this paper uses data sets from microblog network to conduct experiments. The experimental results show that the algorithm effectively combines the homogeneity measure of the network and has achieved very good results in the implicit community discovery.