Research On Agent-based Information Diffusion And Control In Social Networks

With the continuous development of communication and Internet technologies,the scale of social networks expands,and information flows more frequently.Research on information diffusion and control in social networks has theoretical significance and practical value.However,the complexity of the network's strcture,the heterogeneity of nodes and links and the diversity of factors influence the spread all make it difficult to carry out this work.The development of computer technologies and complex networks researches in recent years provides possibilities to study the information diffusion mechanism and control diffusion effects.This thesis uses agend-based modeling method to simulate the information diffusion process and control the diffusion effects on the innovatively proposed social network model.Firstly,technical requirements are extracted from the typical information diffusion scenarios in social networks.Then the social network is modeled based on the analysis of real social network's basic characeristics,including the diversity of nodes,the multiplicity of links and the dynamics of the structure.With the innovative introduction of similarity,a variety of network similarity indexes are given.A social network model based on the similarity is proposed and experimental analysis are carried out.Moreover,social network users are divided into the unknown,listener,spreader and immune and various factors affecting information diffusion are analyzed.A user decision model and a non-linear user decision algorithm are given.An agend-based information diffusion model is construted and experimental analysis are carried out.Finally,kinds of control strategies are proposed based on the analysis of multiple diffusion control factors.Control effects of different strategies on information diffusion are analyzed by conducting experimentsExperiment results show that proposed models can reflect the reality well.Selecting key nodes as initial nodes may speed up the diffusion and expand the diffusion scope,information with higher quality may faster the diffusion speed in the early time and the network with bigger clustering coefficient may make the number of diffusion people fluctuating more intensely.Using influence immune strategy may suppress the information diffusion effectively.