Research On Information Diffusion Model Over Social Networks Based On Evolutionary Game Theory

Combined the social network with the evolutionary game theory(EGT),we model the dynamics of information diffusion over networks and explore the influence of the evolutionary game process under different updating rules and dynamic network structures to the information diffusion.We obtain the expected payoff by applying the mean-field method to describe the star-like relevance between the central participant and his friends.Furthermore,we gain the approximate solution of the evolutionary stable state(ESS)through analytic method.Meanwhile,we simulate the network information dissemination process,and the results have a good consistency with our approximation.Concluding our works as follows:(1)Proposing the information diffusion model based on the evolutionary game theory over social networks.We summarize the correspondence between the terminologies in graphical EGT and those in the social network.The interactions between players can often be modeled by using game theory,while the transmission of information can be regarded as the spread of the gene mutation.Hence,the EGT originated from the Biological evolution theory is very suitable to model the interactive learning and decision process.(2)Building the static information diffusion model over social networks.Under the static network structure,we describe every class of users with his neighbours through mean-field method under different strategy updating rules,including birth-death(BD),death-birth(DB),imitation(IM)and pairwise comparison(PC).The learning and imitation behavior of the central user and his neighbours have been simulated and the analytical expression of the fitness of each class of users under different strategies has been achieved.Specifically,we analyze the framework in uniform degree and non-uniform degree networks and derive the closed-form expressions of the evolutionary stable network states,which are consistent with the simulation results.(3)Building the co-evolutionary information diffusion model over social networks.Based on the static information diffusion model,the two classes of dynamics have been proposed,including the dynamic “of the network” and “on the network”.We adopt three kinds of dynamic mechanism,including the correlation increasing mechanism based on “ triadic closure”,the “useless friends” eliminating mechanism,as well as adding new users and "zombie users" exiting mechanism.The experiment shows some special dissemination phenomenons,which have been discussed and analyzed based on the synthetic networks.(4)Using real case data to analyze and verify the model established in this paper.The real-world Facebook network has been obtained to verify both the static and the co-evolutionary information diffusion model.Experiments show that the proposed framework is effective and practical in modeling the social network users' information forwarding behaviors.