Agent-based Modeling Of Rumor Diffusion In Social Networks

Research on rumor diffusion in social network has great theoretical significance and practical value. However, the complexity of social network, openness of rumor diffusion and heterogeneity between individuals make this study more challenging. Recently, the development of complex network theory and computing experiment methods provide new ideas for rumor diffusion research.Firstly, by analyzing and summarizing the performance of existing different network sampling algorithms in complex networks, a new sampling algorithm is put forward for general social network, based on which the web crawler technology are used to sample the rumor diffusion network. Secondly, typical rumor diffusion scenarios in social network are exampled to extract the general rules of rumor diffusion in social network, then, agent-based modeling is utilized to model the diffusion of rumor in social network. In the process of modeling, agent behavior is divided into three steps, which is recognition, decision-making and action. Moreover, individual psychology and cognition, features of rumor information, behaviors of mass media are considered in the research of rumor diffusion. Finally, the validity of agent-based model is demonstrated and the propagation rules are studied.Experiment results show that agents with larger out-degree have a great influence on the attitude of public and as the initial nodes can significantly improve the diffusion rate. The network clustering coefficient, media authority, coverage frequency of event by mass media and information quality can be different degrees of influence on rumor spreading.