Discrete Opinion Dynamics with Social Bots on Signed Network

Individuals' opinions are affected by their friends or enemies and the relationship can be abstracted as a signed network. However, the emergence of social bots can bring adverse effect to people on the signed network, like misleading the public opinion. In this paper, in order to study the mechanism by which social bots can influence public opinion, we proposed a opinion dynamics model of signed network to study the how the opinion dynamics evolves based on multi-agent model. Afterwards, we give the condition when the probabilities of individuals' selection of opinions will converge in this model. At last, we simulated to show the convergence of our model, and study the effect of different number of social bots and confidence level to the public opinion. The results shows with the increase of the number of social bots, the public opinion is lead out of balance more severely, and as the proportion of social bots reaches 20%, the public opinion of human agents are mislead, and the opinion held by social bots will be the final opinion in the human population. For the population of human agents and social bots, 17% social bots is needed to lead the public opinion. The value of self-confidence has no effect on the final opinion in the signed network, but only whether people are affected by others has influence on the final opinion.