Overlapping Community Detection in Static and Dynamic Networks: A Qualitative Assessment

Community detection (or graph clustering) is one of the fundamental graph analytic techniques for numerous graph-theoretic applications. Many community detection methods have been developed during the last few decades to reveal the concealed properties of complex networks. There are a good number of methods are available for detecting overlapping communities too. It is therefore important to access the performance of the different overlapping community detectors, empirically. More specifically, how good they are in detecting overlapping communities in both static and dynamic networks. In this paper, we present a study on a range of different overlapping community detection methods and access them experimentally to understand better the performance and biases of such methods. We employ six (06) real-world networks to test five (05) state-of-the-art algorithms. We quantify the performances using four (04) different statistical assessment parameters considering both static and dynamic networks. Our experiments reveal that SLPA and SLPAD are the best performers in static and dynamic scenarios, respectively, based on various assessment parameters.