The Statistical Characteristics Of Opinion Dynamics

The evolution of public opinion is influenced by factors such as society, economy. It presents the complexity, openness and non-equilibrium. Therefore, the study of public opinion is more complexity. But, the existing models to study the evolution of public opinion are still inadequate. In this paper, the statistical characteristics of opinion evolution are studied by the knowledge of the complex network theory and statistics. Based on ER networks, WS networks and BA networks, study the importance of individuals in public opinion networks and the influence of trust, influence threshold, government intervention. The main research contents and conclusions are summarized as follows:Firstly, study the importance of individuals(network nodes). In order to improve the study of public opinion evolution based on un-weighted and undirected networks, consider individuals having different importance in the public opinion spreading. Based on degree centrality, betweenness centrality and close degree centrality measures, establish the TOPSIS evaluation method of individuals(network nodes). The evolution of public opinion is generalized to weighted networks. Using TOPSIS method, in ER, WS and BA networks, consider the importance of the individuals. The experimental results show that, in weighted networks, TOPSIS method is more accurate than single evaluation method.Secondly, establish the public opinion evolution model based on trust. Considering the continuous public opinion models lack the study of trust and influence threshold, to introduce the trust definition method, determine the information intensity to change the threshold based on complex network theory. The public opinion evolution model is established on trust. The results show that: with the improving of the impact threshold, the public opinion evolution is showed a strong convergence. The bigger mean is more conducive to reach consensus. The promoting of variance to the evolution of public opinion is related to threshold average position. When the network size is large, the effect of variance is not so obvious. The experimental results, to some extent, explain the social public opinion evolution.Thirdly, establish a dynamic evolution model of public opinion. The network structure, the role of the government, media and others are the important factors to influence the evolution of public opinion. In order to study the neighbor selection, government intervention and network structure, based on complex network theory, the public opinion model of dynamic network is established. The experimental results show that: the greater the neighbor selection proportion is, the more the nodes in the view of government is. When the selection proportion is maximizing, all individual opinion will agree in evolution. Under the government intervention, the public opinion of the network will close to the government public opinion. With the improving of the government impact and influence scope, the public opinion evolution will go faster. Network structure will change as the strength is increased. The network degree distribution will disobey the power-law distribution. The network structure presents a condensation phenomenon. The experimental results, to some extent, explain the public opinion evolution of the society.