Particle Swarm as a Model for Community Formation in Social Networks

Abstract

Online communities play an important role in the world of contemporary social networks. Due to the ubiquitous availability of ICT tools individuals can easily form small, tight, dense groups around various topics and subjects of interest. These online communities share many similarities with general social networks, but they display some interesting distinguishing characteristics. In particular, online communities are usually subject-centered, require some degree of supervision and maintenance in order to continue operations, and provide higher quality user-generated content as compared with larger social networks. Many different approaches have been proposed in the past regarding methods of online community analysis and modeling. Agent-based methods have been successfully employed to simulate online community formation and evolution. However, simulation-based methods inherently suffer from the arbitrary choice of parameters that may produce radically different online community structures as the result of minuscule changes of input parameters. In this paper we examine the usability of a particular agent-based method, namely, the particle swarm model, for online community analysis. We propose a simple model that uses a swarm of particles paradigm to simulate the behavior of users and the formation of online communities and we measure the influence that changing of the swarm parameters has on the resulting online community structure.