MDPCluster: a swarm-based community detection algorithm in large-scale graphs

Abstract

Social network analysis has become an important topic for researchers in sociology and computer science. Similarities among individuals form communities as the basic constitutions of social networks. Regarding the importance of communities, community detection is a fundamental step in the study of social networks typically modeled as large-scale graphs. Detecting communities in such large-scale graphs which generally suffers from the curse of dimensionality is the main objective followed in this study. An efficient modularity-based community detection algorithm called MDPCluster is introduced in order to detect communities in large-scale graphs in a timely manner. To address the high dimensionality problem, first, a Louvain-based algorithm is utilized by MDPCluster to distinguish initial communities as super-nodes and then a Modified Discrete Particle Swarm Optimization algorithm, called MDPSO is leveraged to detect communities through maximizing modularity measure. MDPSO discretizes Particle Swarm Optimization using the idea of transmission tendency and also escapes from premature convergence thereby a mutation operator inspired by Genetic Algorithm. To evaluate the proposed method, six standard datasets, i.e., American College Football, Books about US Politics, Amazon Product Co-purchasing, DBLP, GR-QC and HEP-TH have been employed. The first two are known as synthetic datasets whereas the rest are real-world datasets. In comparison to eight state-of-the-art algorithms, i.e., Stationary Genetic Algorithm, Generational Genetic Algorithm, Simulated Annealing-Stationary Genetic Algorithm, Simulated Annealing-Generational Genetic Algorithm, Grivan–Newman, Danon and Label Propagation Algorithm, the results indicate the superiorities of MDCluster in terms of modularity, Normalized Mutual Information and execution time as well.

Appendix

Figure 13 shows the pseudocode of PSO algorithm. In this algorithm, the main loop iterates as many times as total iterations or as it is known as T. In this loop, there is an inner loop which runs as many times as the number of individuals in the population, N_p, in which velocities and positions are updated. The fitness function is calculated for all the individuals of the population as many times as the number of problem dimensions or |V|. The time complexity of calculating fitness function is $O\left( {\left| V \right|^{2} } \right)$ in this paper. Therefore, the time complexity of PSO is $O\left( {\left( {\left| V \right|*N_{p} + \left| V \right|^{2} *N_{p} } \right)*T} \right)$, and with some simplification, the time complexity of the algorithm is $O\left( {\left( {\left| V \right|^{2} *N_{p} } \right)*T} \right)$.

In GA, the main loop iterates as many times as all the iterations T. in this loop, there is an inner loop which runs as many times as the number of individuals in population N_p in which the fitness function for all individual is calculated and mutation and cross-over are applied on population by mutation and cross-over rates. The time complexity of fitness function in this problem (modularity) is $O\left( {\left| V \right|^{2} } \right)$. Thus, time complexity of GA would be $O\left( {\left( {\left| V \right|^{2} *N_{p} } \right)*T*(p_{c} *O\left( {cross over} \right) + p_{m} *O\left( {mutation} \right))} \right)$. With some simplifications we have $O\left( {\left( {\left| V \right|^{2} *N_{p} } \right)*T} \right)$ for the time complexity of the algorithm.

See Fig. 17.

Fig. 17

Particle swarm optimization