CDCNAs: A comparative study of community detection algorithms in complex networks

Abstract

With the increase in sophistication of diverse network types, a plethora of community detection algorithms in complex networks (CDCNAs) had been introduced. However, a majority of existing surveys and studies are restricted to detecting communities based on selected classes of algorithms. To address this research gap, we briefly present a review of various CDCNAs that includes an overarching taxonomy of these algorithms supported by their in-depth categorization. Furthermore, in pursuit of comparing our study with existing works, we list multiple aspects under which the comparisons are made, suggesting an improvement by our taxonomy. Towards the survey, we classify the CDCNAs into four categories: clusteringbased, nature-inspired, learning-based, and probabilistic graphical model-based. Under each of these categories, the CDCNAs are further divided into one or more classes depending on the properties of the underlying algorithms. A comprehensive version of this survey (in preparation) would further highlight the essence of these contributions.