Influence maximization in social networks using discrete spider monkey optimization

Influence maximization is a classic optimization problem in social network analysis, which plays an important role in product marketing and rumors prevention. Its goal is to select a certain number of the most influential users from a social network as the initial adoptors of information, so that under a specified social network model, the information can achieve maximum influence dissemination. Even though greedy-based methods offer high influence spread, they take long time in large networks, while other existing methods present less satisfactory in effectiveness. In this paper, we propose a discrete spider monkey optimization algorithm for influence maximization and devise new evolutionary rules. In the evolution process, a new individual initialization method and two different node replacement strategies are presented to speed up convergence and reduce running time. Moreover, to avoid the decay of the best individual’s information, we introduce an intercept-keep method. The experimental results on six real-world networks show that our method achieves a performance difference of approximately 1% compared to the state-of-the-art greedy method with less time, and it increases expected dissemination size by 10% to 28% than the effective heuristic and existing swarm intelligence optimization algorithms, indicating the effectiveness and efficiency of our algorithm in addressing the influence maximization problem.