The Research On The Evolution Mechanism Of Collective Behavior In Social Network

The networking of human society just like a coin has two sides,on one side,it improves the level of production and the quality of life;on the other side,it brings some negative effects.Due to the irrational allocation of the resource,the unfairness of income distribution and the obvious gap between rich and poor,social conflicts and contradictions have been intensified,which leading a series of collective behaviors.These events make a huge influence on the public management,the social stability and the security of our country.And there is a common reason behind these events,namely the transitivity of user‘s behavior,which is a typical symbol of human society‘s process.Collective behaviors will easily predict because of the rise of social network and the quantity of users who joined in.So,the study of the evolution mechanism of collective behaviors in social network can help understanding the motives of people in an objective angle,and has both practical value and theoretical significance for the marketing and public opinion control.Based on the related theory in complex network,and combined with the structure features of social network and the characteristics of user behavior,this thesis has built a collective behavior evolution model in social network to tries to explore the internal motivation which trigger users to follow others‘ behavior,the main work is as follows:1.In order to study the inner driven of collective behavior of the complex networks,this paper proposed a collective behavior evolution model based on strategy diffusion,it first studied and analyzed the factor which influence the node making decision and defined three impact indicators from different angles: environmental forces,the difference of strategy,strategies‘ authority,then use the Topsis method to evaluation the three indicate above and get the optimal strategy to update.To evaluate the evolution of the collective behavior,strategy stability,dissimilarity of the system and local agreeableness are used.Finally,experiments are made on real network data,the results show that the propose model not only feasible,but also can well describe convergence properties of the collective in the process of strategy diffusion.2.Based on the analysis of real dataset,several major types of collective behavior in different granular including the node level,the neighbor level,the community level and the social level have been defined.Then a multi-granularity collective behavior diffusion framework is proposed to quantize the inner driven of collective behavior.The experimental results verify the existence of the conformity of collectives.Leveraging the infectious diseases model,collective behavior can be described from different levels,the experiments results demonstrate that individuals can rationally make their decision when they are isolate with others.But when neighbor exist,their actions will be irrational,the phenomenon will be more obvious when social reinforcement existence.Members in the same community can easily reach a group consensus while different communities hold different opinions.