Selection strategy in graph-based spreading dynamics with limited capacity

Fei Xiong, Yu Zheng, Weiping Ding, Hao Wang, Xinyi Wang, Hongshu Chen

Abstract: Highlights • We study effective strategies of altering epidemic spreading with limited capacity. • We propose a graph-based diffusion model in which spreaders only contact a finite number of neighbors. • Selection strategies change the diffusion extent but do not alter the spreading threshold. • The diffusion benefits from the preference of small betweenness and clustering coefficients rather than large descriptors. Abstract Recent studies revealed that node similarities which characterize common links between nodes induce structural redundancy, and large redundancy is not effective for the diffusion in social networks. The phenomenon was verified in the context of independent cascades. In this paper, we concentrate on effective strategies of altering epidemic spreading in consideration of limited capacity. We propose a new diffusion model in which spreaders only contact and infect a finite number of neighboring nodes. Different strategies are taken by spreaders to select neighbors as contact targets. We further investigate the roles of selection strategies in the dynamics. Analytical and simulation results in artificial graphs prove that selection strategies change the final diffusion extent but do not alter the spreading threshold. Phase transition depends on the spreading rate and the number of contact targets. Contrary to independent cascades, selecting nodes with large similarities preferentially promotes the diffusion most effectively in epidemic dynamics with limited capacity. Dramatically, the diffusion benefits from the preference of small betweenness and clustering coefficients rather than large descriptors. Dynamics in real-world networks confirms the analytical results.