DARIM: Dynamic Approach for Rumor Influence Minimization in Online Social Networks

This paper investigates the problem of rumor influence minimization in online social networks (OSNs). Over the years, researchers have proposed strategies to diminish the influence of rumor mainly divided into two well-known methods, namely the anti-rumor campaign strategy and the blocking nodes strategy. Although these strategies have proven to be efficient in different scenarios, their gaps remain in other situations. Therefore, we introduce in this work the dynamic approach for rumor influence minimization (DARIM) that aims to overcome these shortcomings and exploit their advantage. The objective is to find a compromise between the blocking nodes and anti-rumor campaign strategies that minimize the most the influence of a rumor. Accordingly, we present a solution formulated from the perspective of a network inference problem by exploiting the survival theory. Thus, we introduce a greedy algorithm based on the likelihood principle. Since the problem is NP-hard, we prove the objective function is submodular and monotone and provide an approximation within $$(1-1/\textit{e})$$ of the optimal solution. Experiments performed in real multiplex and single OSNs provide evidence about the performance of the proposed algorithm compared the work of literature.