STUDYING THE EFFECTIVENESS OF COMMUNITY DETECTION ALGORITHMS USING SOCIAL NETWORKS

Abstract

Social network analysis is a significant area of research for analyzing the interconnection between the people within network. Community detection is one of the most important applications in SNA. The main motive of CD is to discover the collection of node that are tightly correlated within the network and weakly correlated to another network for partitioning the network to form the group of communities. The aim of this work is to detect communities from undirected disjoint social networks in which it is implemented on lesmis and email-Eu-core-department-labels networks. Effective partitioning and detection of the network are the primary factors for implementing this work by using Girvan–Newman, greedy modularity maximization, and Kernighan–Lin bipartition CD algorithms. The effectiveness of these CD algorithms is analyzed with respect to ground-truth communities based on measures such as recall, normalized mutual information score, precision, and F1-score. Experimental results show that the greedy modularity maximization algorithm provides best results for CD on email-Eu-core-department-labels network with respect to corresponding ground-truth communities.