Dynamic mechanism of social bots interfering with public opinion in network

Participants in discussions on online social networks tend to become polarized into clusters of users with diametrically opposite opinions. Recent evidence has suggested that social bots are being used on social media networks to manipulate public opinion, but this mechanism has not been adequately investigated. In this paper, using the spiral of silence theory of social communication, we establish a multi-agent model based on user interactions on social media, define the behavioral characteristics of social bots and human users at the microscopic level, and reveal the mechanism of manipulation of public opinion by bots. The results of simulations of small-world and scale-free networks show that social bots need only constitute 5%-10% of participants in a given discussion to alter public opinion such that the view being propagated by them eventually becomes the dominant opinion (held by more than 2/3 of the population). The influence of network density, efficiency of clustering, and spatial location on the manipulative effect of bots was analyzed. The results show that social bots can influence the formation of opinions on online social networks. (C) 2020 Elsevier B.V. All rights reserved.