Multiobjective Overlapping Community Detection Algorithms Using Granular Computing

Community detection is one of the most important problems in Social Network Analysis. This problem has been successfully addressed through Multi-objective optimization Evolutionary Algorithms (MOEAs); however, most of the MOEAs proposed only detect disjoint communities, although it has been shown that in most real-world networks nodes may belong to multiple communities. In this chapter, we introduce three algorithms which build, from different perspectives, a set of overlapping communities using Granular Computing theory and based on a Multi-objective Optimization approach. The proposed algorithms use highly cohesive granules as initial expansion seeds and they employ the local properties of the vertices in order to obtain well accurate overlapping communities structures.