Research On Social Network Population Event Dynamics Based On User Behavior

In the analysis of social network public opinion,research on the diffusion mechanism of group events and hot topics in the network has become a hot spot of current research.As the interdisciplinary field of online social network analysis and complex network research,information diffusion dynamic forms a huge diffusion network,which integrates user relationship and information interaction.Meanwhile,the complexity and multidimensional evolution of online social network structure,as well as the diversity of user behavior characteristics,bring difficulties and challenges to the analysis of event diffusion in social network groups.How to accurately analyze the dynamic causes of social network group events,explore the mechanism of information diffusion,and excavate the main features are the current problems to be solved.This thesis of information diffusion research mainly focuses on two aspects.For the study of single message,the information diffusion trend is analyzed on the basis of multidimensional network and game theory,and the dynamics of social network group events is excavated.For the study of multiple message,the dynamic interactive diffusion process of multiple message is analyzed on the basis of multiplex networks and timedependent variables,and the trend of multiple message diffusion is explored.The main contents of this thesis are listed as follows:1.Dynamics research around single message and multidimensional networks.Aiming at the heterogeneity of propagation path and the complexity of user interaction,a user behavior evolution strategy based on topic heat is proposed.The purpose is to explore the impact of static and dynamic driving factors on group state transitions during information diffusion.First,user attributes and interaction attributes are extracted,and mapped into multidimensional network space: behavior influence,attribute influence,and topological influence subnets,so as to reduce the coupling relationship between subnets.Second,assuming that the psychological characteristics of topic user are the driving factors affecting information diffusion,the concept of topic heat is defined.On the basis of evolutionary game theory,a dynamic evolution strategy of user behavior is proposed.Finally,on the basis of the traditional epidemic model,the improved information diffusion dynamic model is obtained by combining the multidimensional network and psychological characteristics.2.Furthermore,the thesis extends the single message and multidimensional networks to the dynamic research of multiple message and multiplex networks.Aiming at the interaction among multiple message diffusion process and display rich intertwined effects,a diffusion model based on multiple message and multiplex network space is proposed,which aims to explore the internal mechanism of multiple message concurrent diffusion under social topics.First,aiming at the multidimensional characteristics of diffusion space,the overlap of social structure and the complexity of diffusion behavior are mined and mapped to multiplex diffusion space.Through hierarchical processing,not only multidimensional message diffusion path can be constructed,but also the diffusion situation of each message can be further perceived.Second,for the problem that the infectious rate depends on time change in information diffusion,the time-dependent variables of interaction between multiple message is introduced.In this way,the diffusion process of cooperative and competitive between multiple message can be reflected.Finally,the dynamic interaction mechanism is introduced based on the multiplex epidemic model.A unified framework of multiple message coupled dynamic processes is constructed by discrete-time microscopic Markov chain approach.At last,the thesis validates the proposed model using real and simulation data.Experiments show that the proposed model is feasible and effective.The model can not only effectively reveal the influence of different driving factors on information diffusion,but also perceive the information diffusion trend in social networks,and more truly reveal the multiple message interaction mechanism under social topics.