The Social Networks Evolution Model Based On Hypermap Algorithm And Research On Diffusion Source Localization Method

In social networks, all kinks of rumors always propagate, which awfully threaten the stability of country and society, so it is of great significance for predicting diffusion range and controlling diffusion procedure to effectively localize information diffusion source. Social networks’ main feature is dynamic feature, that is, over time, in social networks, the nodes either increase or decrease and the edges either increase or decrease. Therefore, it is import for localizing information source to be able to greatly simulate social networks evolution law by modelling the social networks evolution. The paper is based on two preconditions:the first is we assume the social networks evolution proceeds just from edges increasing perspective; the second is we assume that we know the current diffusion topology and the difference between the time of localization and the time of formation of the real diffusion. The greatest difference between this paper and previous localization algorithms is considering social networks’ dynamic evolution, thus compared with the Extended Single Source Localization Algorithm Based on Observers, the paper makes relative improvement in localization performance.Starting with social networks’ dynamic feature, and taking account of the topology which is at the time when we localize the information source is different from the real diffusion topology, the paper apply EPSO model to model social networks and predict links in social networks by HyperMap Algorithm. Applying HyperMap Algorithm, We predicted new generated links based on the current diffusion topology, and gain the real diffusion topology by deleting the predicted links from the current diffusion topology. Lastly, on the basis of that we apply single source localization algorithm localize the information source.The paper dopts HyperMap Algorithm to predict future links in synthetic and real networks,whose diffusion model is random diffusion model, which is SI model. At the time of localize the information source, we make a variety of contrast experiments, for example, at the same diffusion topology, different observers deployment policies; at the same diffusion topology, different observers deployment proportion. From the results of experiments, in general, our localization algorithm performance is superior to the Extended Single Source Localization Algorithm Based on Observers’.Therefore, we can conclude that the algorithm that the paper put forward has a remarkable effect on localizing source in social networks and which is distinctly import for localizing and controlling rumors in social networks.