Detecting communities in topical semantic networks

Abstract

With the advance of information technology, online communications between people have increased significantly. These kinds of communications have become more organized subsequent to the emergence of social networks. One of the most important issues in analyzing these kinds of networks is community detection, in which most studies detect communities through analyzing linkages of the networks. The desirable goal in this paper is to reach communities in which the members have the same topic of interest, and where the strength of connections between them is the consequence of their communications' content analysis. Therefore, we propose a new framework which considers the information that is shared by the users, and also the topics they are interested in, in order to find more meaningful communities. While similar studies have only found communities by merely considering the topological structure of the network, and some of the features of semantic information related to the users of the network, like their topics of interest, the proposed framework detects communities through considering topics, communications' content and topological structure of the network. Quantitative evaluations indicate that the proposed framework achieves promising results which are superior in comparison with the other relevant frameworks in the literature.