Influence spreading model used to analyse social networks and detect sub-communities

Abstract

A dynamic influence spreading model is presented for computing network centrality and betweenness measures. Network topology, and possible directed connections and unequal weights of nodes and links, are essential features of the model. The same influence spreading model is used for community detection in social networks and for analysis of network structures. Weaker connections give rise to more sub-communities whereas stronger ties increase the cohesion of a community. The validity of the method is demonstrated with different social networks. Our model takes into account different paths between nodes in the network structure. The dependency of different paths having common links at the beginning of their paths makes the model more realistic compared to classical structural, simulation and random walk models. The influence of all nodes in a network has not been satisfactorily understood. Existing models may underestimate the spreading power of interconnected peripheral nodes as initiators of dynamic processes in social, biological and technical networks.