Minimizing Rumor Influence in Multiplex Online Social Networks Based on Human Individual and Social Behaviors.
Adil Imad Eddine Hosni,Kan Li,Sadique Ahmad
DOI: https://doi.org/10.1016/j.ins.2019.10.063
IF: 8.1
2020-01-01
Information Sciences
Abstract:With the growing popularity of online social networks, an environment has been set up that can spread rumors in a faster and wider manner than ever before, which can have widespread repercussions on society. Nowadays, individuals are joining multiple online social networks and rumors simultaneously propagating amongst them, thereby creating a new dimension to the problem of rumor propagation. Motivated by these facts, this paper attempts to address the rumor influence minimization in multiplex online social networks. In this work, we consider modeling the propagation process of such fictitious information as a significant step toward minimizing its influence. Thus, we analyze the individual and social behaviors in social networks; subsequently, we propose a novel rumor diffusion model, named the HISBmodel. In this model, we propose a formulation of an individual behavior towards a rumor analog to damped harmonic motion. Following this, the opinions of individuals in the propagation process are incorporated. Furthermore, the rules of rumor transmission between individuals in multiplex networks are incorporated by considering individual and social behaviors. Further, we present the HISBmodel propagation process that describes the spread of rumors in multiplex online social networks. Based on this model, we propose a truth campaign strategy in minimizing the influence of rumors in multiplex online social networks from the perspective of network inference and by exploiting the survival theory. This strategy selects the most influential nodes as soon as the rumor is detected and launches a truth campaign to raise awareness against it, so as to prevent the influence of rumors. Accordingly, we propose a greedy algorithm based on the likelihood principle, which guarantees an approximation within 63% of the optimal solution. Systematically, experiments have been conducted on real single networks crawled from Twitter, Facebook, and Slashdot as well as on multiplex networks of real online social networks (Facebook, Twitter, and YouTube). First, the results indicate the HISBmodel can reproduce all the trends of real-world rumor propagation more realistically than the models presented in the literature. Moreover, the simulations illustrate that the proposed model highlights the impact of human factors accurately in accordance with the literature. Second, compared to the methods in the literature, the experiments prove the efficiency of our strategy in minimizing the influence of rumors in the cases of single network and multiplex social network propagation. The results prove that the proposed method can capture the dynamic propagation process of the rumor and select the target nodes more accurately in order to minimize the influence of rumors.