Label entropy‐based cooperative particle swarm optimization algorithm for dynamic overlapping community detection in complex networks

• DOI: 10.1002/int.22673
• Publication Date: 2021-09-24T22:58:54Z
• Authors: Wenchao Jiang; Shucan Pan; Chaohai Lu; Zhiming Zhao; Sui Lin; Meng Xiong; Zhongtang He
• Keywords: 0202 electrical engineering, electronic engineering, information engineering; 02 engineering and technology; 510; 004

The real-world complex networks, such as biological, transportation, biomedical, web, and social are usually dynamic change over time. communities which reflect the substructures hidden in networks overlap each other, detecting overlapping is a challenging task. Prior researchers have applied multiobjective optimization method to detection of achieved some excellent results. However, terms processing, prior studies all adopt decomposition based on weight parameters, different parameters or parameter values can easily affect community results further uneven distribution detected target space. To solve above problems, hybrid algorithm, that is, Collaborative Particle Swarm Optimization-based Dynamic Overlapping Community Detection (CPSO-DOCD) algorithm proposed this paper. First, improve diversity particles, encoding/decoding particle cross inheritance variation redefined first label propagation. In network snapshot, multiple swarms initialized Overlap Propagation Algorithm (COPRA) generate particles with uniform distribution. Multiple objective functions optimized using respectively avoid incorrect selection parameters. addition, reference-point-based adopted selecting stage Second, node entropy-based swarm accuracy current snapshots. Finally, when one snapshot switches another time, migration strategy COPRA local-search clique generation utilized adjust results, enables former be adapted new experiments implemented four Cit-HepPh, Cit-HepTh, Emailed-EU-core-temporal, CollegeMsg. hypervolume value result obtained by CPSO-DOCD 0.5%–2% higher than MDOA, MCMOEA, SLPAD, iLCD. Furthermore, also performed better iLCD C-metric values, approach approximately Pareto frontier.