Analysis of Different Measures of Centrality to Identify Vital Nodes in Social Networks

Social networks are essential for connecting individuals from around the world. Finding the vital nodes inside a social network is still difficult because of the different sizes of the network. Numerous measures of centrality have been created in order to address this problem. Finding vital nodes is crucial for activities like disease management, increasing product exposure, promoting or blocking the dissemination of information, and many others. In this paper, we analyze some earlier and recently proposed measures of centrality, including degree, betweenness, eigenvector, closeness, K-shell decomposition, gravity index, extended gravity index, local gravity model, K-shell gravity centrality, and global and local information, in order to determine the best centrality measure in terms of accuracy, distinguishing ability, and efficiency. Therefore, different experiments are executed on real networks using the ten different centralities mentioned above. The SIR model and Kendall tau are employed as evaluation criteria to evaluate the performance of these ten measures of centrality. The empirical outcomes demonstrate that the extended gravity index centrality has better accuracy and distinguishing ability as compared to other centralities. At last, we present our future research strategies based on centralities.