Research On The Evolution Of Continuous Group Opinions In Social Networks

Recently the issue of the evolution of group opinions has attracted much attention of scholars.To reveal the relationship between micro-level characteristics of individuals and the macro-level rule of opinion evolution is very important.This issue can be widely used decision-making field,such as designing the process of decision-making for a large-scale group of people more effectively.Furthermore,the study of this issue will also help us to guide the spread of public opinions well,so that the social group's opinions can conform to the mainstream values of our country.This dissertation mainly discusses the basic laws and the factors of the evolution of the group's continuous opinions in social networks.The general idea is modeling-analysis-simulations.The main work and innovation of the dissertation include the following aspects:Research Content 1:Modeling and Model AnalysisIn order to be better consistence with the laws of social psychology,this dissertation proposes a new model of opinion evolution based on social judgment theory and the classical Hegselmann-Krause model.Then the relevant properties of the model(global interaction and local interaction)are proved.Under global interaction,the evolution of the group opinions will follow the rules of order preservation and mean invariance.Under local interaction,if trust is mutual,the individual has self-confidence,and the influence between individuals has a small lower bound,the model will convergent.Research Content 2:Explore the impact of social network topology on the evolution of community perspectives.In order to quantify the influence of various factors on the evolution of group opinions accurately,this dissertation uses the indicators-group opinions distance function.On the fully-connected graph,it is found that the two factors-l and k have an significant effect on the evolution of group opinions,where the parameter l reflects the individual openness and the parameter k reflects the individual fluctuating performance.In addition,it is found that the impact of interaction patterns on the evolution of group opinions is more important than the influence force between individuals and individual's self-confidence.It is also found that the topology of small-world social networks(the edge-adding probability p and the number of nearest neighbors K)significantly affect the convergence time of the group opinion and the number of clusters.Using the Holme-Kim's scale-free network generation model,I use information entropy to measure the dispersion of group opinions.And it is also shown:to reach a consensus on a scale-free network is much more difficult than other social network structure.The influence of community structure on the evolution of group opinions was discussed.It is proved that if individuals have self-confidence,trusts are mutual,and the underline graph is connected,the group opinions will convergent.When the connections between subgroups are infinitely small,the convergence time of group opinions will tend to be infinite.We find that increasing the number of connections between subgroups can accelerate the convergence of group opinions,while increasing the number of subgroups within the subgroups slows the convergence rate.Research content 3:The influence of the heterogeneity of group members on the evolution of group opinions.In this dissertation,the effect of the heterogeneity of individual openness on evolution is studied on full-connected graph and scale-free network respectively.To study the effects of different proportions of close-mind individuals and open-mind individuals on evolution,ANOVA is used to compare the mean effects when the network structure is fully-connected.In order to reflect the influence of the changes of different types of groups on the evolution of group perspectives in more detail,this dissertation studies the influence of heterogeneity by means of changing the distribution of individual openness.It is found that without changing the lowest openness of individuals does not improve the group's ability to aggregate.The effect of changing the lower bound of individual's openness is much better than changing the mode of the distribution.When the network structure is scale-free,a method of detecting influential nodes based on opinion dynamics is proposed.Then the effect of positive matching(high-impact nodes have high openness)and reverse matching are compared.Results show that the degree of agreement of the group's opinions is much higher than that of the other two cases(reverse matching and homogeneity).In addition,it is found that the higher the heterogeneity,the easier the group's opinion can aggregate.