Parametric Classification of Dynamic Community Detection Techniques

Abstract

The community detection in a given network is the idea to find a cluster in the structure. A community is the most densely populated part of the graph. The observed network is mostly sparse having multiple dense partitions in it, for example, a protein–protein interaction network where different proteins interact with each other. Here, communities can be detected by finding the cluster of proteins in the network to find different functional modules. Another example is of Facebook friendship network. Several authors try to find the structure and communities in this type of network. Multiple clusters in one network can also be detected which can overlap with each other. This paper covers the classification of different community detection techniques in dynamic networks and then compares them on the basis of different features, e.g., parallelization, network models, community instability, temporal smoothness, etc.