Analysis And Modeling Of Public Opinion Dissemination Based On Complex Network

Public opinion,as a complex and related network,is highly similar to a complex network structure,so complex networks have gradually become an important means of researching public opinion networks.Based on the characteristics of public opinion,there are many ways to form,fast propagation,and wide spread.When emergencies occur,public opinion will spread quickly.If the relevant departments do not handle it properly at this time,it may cause a crisis of public opinion and even affect social stability.Research on the analysis and modeling of public opinion communication in complex networks is helpful for analyzing network characteristics,mining network structure,and predicting network development.It has important theoretical research value and practical significance for monitoring and correctly guiding network public opinion.Study the laws and influencing factors of information dissemination on social network is of great significance for analyzing the spread of public opinion,preventing the spread of rumors and guiding information transmission.This thesis improves the shortcomings of the traditional Susceptible-Exposed-Infective-Removed(SEIR)model.We establish the Susceptible-Exposed-Trusted-Question-Recovered(SETQR)model and use the probability theorem to derive the law of information propagation.At the same time,the equilibrium point and the basic regeneration number of the SETQR model are solved by differential dynamics and regenerative matrix method.The stability of the SETQR model at the equilibrium point is proved by theoretical derivation.Finally,the experimental verification is carried out.The simulation results show that the SETQR model has local stability at the equilibrium point,which is consistent with the theoretical analysis.Through further simulation,the effects of time-delay mechanism,containment mechanism and forgetting mechanism on the speed of information dissemination and the time required for the network to reach equilibrium are analyzed.It is significant to study the community discovery algorithms for complex networks for digging out the origin of opinion,analyzing the ways of public opinion transmission and controlling the evolution of public opinion.To solve the problem that the existing clustering algorithm of central node have low quality of community detection.In this thesis,it proposes a community detection method based on Two-layer dissimilarity of central node(TDCNCD).The algorithm selects the central node through the degree and distance of the node at first.It is avoided that nodes in the same community that are close are selected as the central node at the same time.At the same time,the algorithm proposes the dissimilarity index of nodes based on two layers,which can explore the heterogeneity of nodes deeply and achieve the effect of accurate community division.Simulation experiments are carried out in three real data sets.The results show that compared to the classical community partitioning algorithms,the TDCN-CD algorithm can detect the community structure effectively and divide the community more accurately.It is important to study link prediction algorithms for complex networks to analyze the direction of public opinion communication,predicting the trend of public opinion evolution,and controlling the development of public opinion.In this thesis,after investigating existing link prediction algorithms,a link prediction algorithm combining two-layer degree of code and clustering coefficient(TDNCC)is proposed to further improve the accuracy of the algorithm.The algorithm comprehensively considers the local structure information of the network and the difference between the common neighbor nodes.The node degree and the clustering coefficient are combined in the selection of the similarity evaluation index,and the similarity property of the deep mining node extends the node degree to two layers.Finally,simulation experiments are carried out in three real data sets.The results show that the proposed algorithm has better performance than classical algorithms such as Common Neighbors,Adamic-Adar and Resource Allocation.