Abstract

Social Network Analysis (SNA) is focused on discovering and analyzing the pattern of interactions between the nodes of a social network to understand the implications of these interactions. Community Detection divides a social network into meaningful groups with nodes having frequent interactions among them. In order to obtain semantically relevant communities, we need to consider not only the structural information but also semantic information associated with nodes. Identifying such communities have wide applications in viral marketing, recommender systems etc. In this paper, we propose a method which leverages the semantic content along with structural interactions to detect the non overlapping communities which are not only strongly connected but also have similar interests. Experimental study demonstrates our proposed method detects semantically relevant communities as compared to structure based approaches.