Agent-based Modeling And Experimental Analysis Of Information Diffusion In Social Networks

Research on information diffusion has great theoretical significance and practical value in social networks.However,the complexity of the link relationship,the diversity of the dissemination of information,and the heterogeneity of the individual have brought some difficulties to the success of this work.In recent years,the rapid development of computer technology provides a new idea for the research of complex networks,which provides the possibility for the research of information transmission law and the accurate prediction of user behavior.Based on the network structure model,the agent-based modeling method is proposed to simulate the information dissemination process,summarize the propagation rules and predict user's behavior.Firstly,typical information diffusion scenarios in social network are exampled to extract the general law of information diffusion and user decision-making.Secondly,the social network is modeled based on the link vector model.Online social network is the carrier of information dissemination,and portraying the reality of the social network structure can reproduce more real law of the spread.The link relationship between nodes is divided into five categories: kinship,geo-relationship,industry relationship,intimate relationship and marriage relationship in order to analyze the influencing factors of different relationship intensity.Moreover,traffic relation network is introduced to establish the interrelation model of the five links.In the process of modeling,this thesis divides the agent behavior into four steps: cognition,persuasion,decision-making and diffusion,and adopts the rule-based individual activity decision algorithm and the multi-state propagation decision algorithm to judge whether the individual will participate the activity and whether the information will be transmitted.Finally,the validity and correctness of the model are demonstrated.Experiment results show that as publishers agents with larger out-degree can not only enhance the effect of information dissemination,but also have a certain guiding effect on user behavior decision.What's more,information quality,neighbor node's attitude and release frequency of information and a variety of other factors have different effects on the dissemination.It is also discovered that the more cost individual decision takes and the more rational it is,the corresponding individual behavior will be much easier to be predicted.