Community Detection Using Meta-heuristic Approach: Bat Algorithm Variants

In the present world, it is hard to overlook - the omnipresence of 'network'. Be it the study of internet structure, mobile network, protein interactions or social networks, they all religiously emphasizes on network and graph studies. Social network analysis is an emerging field including community detection as its key task. A community in a network, depicts group of nodes in which density of links is high. To find the community structure modularity metric of social network has been used in different optimization approaches like greedy optimization, simulated annealing, extremal optimization, particle swarm optimization and genetic approach. In this paper we have not only introduced modularity metrics but also hamiltonian function (potts model) amalgamated with meta-heuristic optimization approaches of Bat algorithm and Novel Bat algorithm. By utilizing objective functions (modularity and hamiltonian) with modified discrete version of Bat and Novel Bat algorithm we have devised four new variants for community detection. The results obtained across four variants are compared with traditional approaches like Girvan and Newman, fast greedy modularity optimization, Reichardt and Bornholdt, Ronhovde and Nussinov, and spectral clustering. After analyzing the results, we have dwelled upon a promising outcome supporting the modified variants.