Improved centrality indicators to characterize the nodal spreading capability in complex networks

Abstract

In this paper, we deeply investigate the identification of influential spreaders in complex networks based on various centrality indices. At first, we introduce several frequently used centrality indices to characterize the node influence. Then, based on the standard SIR model, we integrate various centrality indicators into the characterization of the nodal spreading capability, and then starting from the gravitational centrality formula, we systematically compare the ranking similarity and monotonicity under various centrality algorithms over 6 real-world networks and Barabási-Albert model networks. The extensive simulations indicate that the mixed measure of gravitational centrality combining the k−shell value and degree will display the best performance as far as the ranking results are concerned, in which the focal node used the k−shell value as his mass while his neighboring nodes viewed the degree value as their masses. The current results are beneficial for us to develop the effective methods to discover and protect the significant nodes within many networked systems.