Sequential seeding strategy for social influence diffusion with improved entropy-based centrality

In this paper, we investigate the centrality problem of selecting seed targets for sequential seeding strategy in social networks. Based on the concept of entropy, we design a novel improved centrality by integrating interaction intimacy and confidence level to measure the total influence of an individual which can be decomposed into direct effect and indirect effect. In addition, we formulate the sequential seeding strategy to evaluate the performance of the proposed centrality and compare it with the counterpart of the single-stage seeding strategy. Furthermore, extensive experiments are conducted for comparison with the other centralities including betweenness, closeness, degree, and eigenvector in two empirical and four artificial social networks. By simulations, we find that the proposed entropy-based centrality is superior to other centralities in terms of diffusion speed and influence coverage in the BA scale-free network. Parameter analysis of sequential seeding strategy demonstrates that the proposed centrality can achieve the greatest total influence coverage in the case where the individual's confidence in each neighbor is treated equally. (C) 2019 Published by Elsevier B.V.