An Interaction Based Influence Propagation Model for Detecting Active Communities in Social Networks

Abstract

Social Network Community Detection (SNCD) is one of the main challenges in social media analytics which is defined as a collection of people who share common interests. Since the volume of data produced everyday by social networks is growing enormously, the challenge is finding active and dense communities demonstrate constant interactions among its members. The existing SNCD techniques had focused mostly on the static nature of the communities and weightage is concerned for the number of connections but in order to understand the real behavior of the communities it is worthwhile to consider interactions among users in a network along with the friendship/followership. In this paper, an Interaction based Influence Propagation Model (IIPM) is proposed which makes use of k-clique which will help in finding dense and active communities. In this work, A Clique Percolation Method is used to identify the number of cliques along with Influence Propagation Model, where the weights are generated for each node in order to include inactive communities if these are influenced by active neighbors. The proposed model considers interactions as weight parameter presuming that interactions can provide information about user’s behavior and can predict of the activities of the user. The effectiveness of the detected communities and the efficiency of the proposed IIPM is verified on benchmark datasets and experimental results have shown the Significance of Interactions in IIPM and outperformed over the existing SNCD Methods.