The Modeling And Analysis Of Opinion Evolution And Decision-making Mechanism Over Social Network

Opinion dynamics is an emerging interdisciplinary research field that combines mathematics,physics,computer and human behavior and studies the phenomenon of group opinion evolution based on mathematical models and simulations.It mainly studies the mechanism of the formation and evolution of individual’s opinions or actions in social groups due to individual characteristics,environmental factors,and policy guidance in communication activities.Through modeling and analysis evolution process of opinions of social groups.This can be used to dig out the internal mechanism that affects the evolution of opinions,and it is also conducive to the promulgation and implementation of some policies and instructions.Scholars have conducted many related researches by using opinion evolution mechanism modeling and multi-agent simulation analysis methods.The bounded confidence model studies the psychological property that people do not listen to individual opinions that contradict their own.The individual’s action is associated with their opinions.Martin combined this external behavior with the individual’s internal motivation(opinion)to study the phenomenen of opinion polarization by using continuous opinion and discrete action(Continuous Opinions And Discrete Actions,CODA)model.Besides,people always have obvious semantic features when expressing their opinions,and this scene can be modeled and analyzed through fuzzy inference rules.To study such a interesting topic,this thesis proposes a variant CODA model by combining fuzzy inference system and bounded confidence rule.The impact of bounded confidence level and social network structural attribute differences on social groups’ opinion evolution and results is studied by using opinion polarization theory,social network theory,and opinion dynamics simulation analysis methods.Compared with the CODA model,the group opinions under the constraints of bounded confidence rule will be updated synchronously,which makes the group opinions show a tendency of polarization faster.When the bounded confidence level increases,the convergence speed of the group’s opinions will increase,and the group will reach a more consistent consensus.Next,this thesis proposes a variant ”time-based small-world network generation algorithm”,and analyzes the influence of social network structure differences on the formation and evolution of group opinions.The relevant simulation results show that the few shortcut edges in the small world network can affect the spread of extreme opinions in the group,and will promote social group reach consensus.Compared with the static smallworld network,the it takes longer for individuals to reach a consensus on the time-based small-world network.However,a higher polarization consensus will reach after longer communication.Finally,this thesis proposes a co-evolution model of group opinions and actions on hierarchical networks.The impact of confidence level on the opinion evolution of social groups over heterogeneous hierarchical networks and homogeneous hierarchical networks were comparatively analyzed.The simulation results show that the group opinions and actions on the hierarchical network will show polarization.However,as the size of the group increases,the efficiency of opinion dissemination in the network will decrease,and the views within the group cannot reach a good consensus.