Information Dissemination Model Based on Clustering Analysis of Information Network Development

Abstract

Our goal is to analyze the way information flows in the social network and to draw a conclusion for the relationship between the information spread and public opinion. First, by utilizing K-Means Algorithm with PCA (Principal Component Analysis), we build a comprehensive assessment system of information development. Second, based on the traditional SIR Model in epidemic field, we build a SIER dissemination model introducing the conception of Super-spreader in disease propagation. Then we analyze the changing trend of different nodes' density influenced by the Super-spreaders. According to the results of clustering, we set different values of parameters in the model respectively matched with each clustered group. In this way, our paper successfully reveals the information dissemination mechanism and showing the dynamic diffusion of information flow in the social network. Furthermore, we validate our model's reliability by comparing the results we predict with the realistic data and predict the situation of communication network successfully. After that we perform sensitivity analysis by studying important factors which impacts the public opinion and information spread change in social network. Finally, this paper proposes some advice for policymakers to control and guide the flow of information via adjusting important factors in the diffusion model.