Community verification with topic modeling

Abstract

Different performance measurement metrics have been proposed to evaluate the performance of community detection algorithms, such as modularity, conductance, etc. However, there is few work which makes sense of a community, that is, explain what does the community do, what is the community’s interest. In this paper, we use topic modeling to capture the topics of users in the same community and verify a heuristic community detection algorithm by showing that the users in the communities share strong interests.