Influence propagation based community detection in complex networks

Abstract

Interaction between nodes in a complex network showing the property of homophily tends to produce community structure in the network. The detection of these communities is of immense financial and informational value. For this purpose, we propose a semi-supervised community detection algorithm, inspired by genetic genealogy and based on Label Propagation Algorithm, that detects communities in the network by taking into account the propagation of influence from different community centers identified in the network. Analysis of our proposed algorithm showed improved performance in detecting communities in real social networks.