On The Propagation Mechanism Of Public Opinions With Bounded Confidence Rule Under Emergencent Collective Events

The occurrence of emergent events is an inevitable phenomenon during the development of the human society. Emergent events often possess some characteristics such as calamity, urgency and uncertainty and easily drive the public into huge panic, which increases the difficulties of the management and control of emergency. At the same time, a series of public opinions always company with the occurence of emergent collective events and influence the opinions and beaviors of the public. From the 21 st century, the propagation stytles and ways of public opinions have had many important changes, thus, building a sound guidance policy for public opinions under emergency has become a commonly concerned and also very urgent problem for all the governments. The thesis, from the point of view of propagation law of public opinions under emergency, aims at exploring the underlying physics laws of the propagation of public opinions and analyzing some factors on the opinion propagation with typical cases in real life by using some theories and methods from social network theory, social dynamics and complex systems science. A more complete research framework is given for modeling and analysis of opinion dynamics, which will provide theoretical evidence and an effective measure for emergency management.Opinion dynamics is an important kind of social dynamics and also a basic approach to build the models of opinion propagation. Some classical opinion dynamics models include discrete opinion models, such as Voter model, Majority rule model, Social impact power model, Sznajd model, and continuous opinion model, such as Deffuant-Weisbuch(DW) bounded confidence model, Hegselmann-Krause(HK) bounded confidence model and their extended versions. Since the objective of the thesis is the propagation of continuous opinions, then bounded confidence rule is applied to study the evolution of the collective opinions. Some opinion dynamics models are built to investigate the evolution law of the collective opinions under emergency.The main content and results of the thesis include the following four points.1. A heterogeneous bounded confidence model is built with a more general form. Analyses are given for the influence of some factors, such as confidence levels and the size of a social network, on the opinion propagation with computer simulations. When the confidence levels of social agents are heterogeneous, the collective opinions are more difficult to reach consensus than homogeneous HK models. The difficulty increases as the differences of the confidence levels increase. For a homogeneous HK model, the size of a social network only effects the convergence time of the opinion dynamics. However, for a heterogeneous HK model, there is a positive correlation between the size of a social network and the number of the final opinion clusters. A heterogeneous bounded confidence rule is more appropriate for real societies, where agents have their own characteristics and provides a new approach to study the opinion dynamics.2. A multi-level heterogeneous bounded confidence model is built with considerations on the heterogeneity and the hierarchy of confidence levels. The social network is divided, according to the individual difference theory, into open-minded subnetwork, middle-minded subnetwork and close-minded subnetwork. The computer simulation results show that the state of the final collective opinions depends on the social agents with the smaller confidence levels. Firstly, when the size of the social network is fixed, there exists a threshold of the fraction of the close-minded agents. If the fraction of the close-minded agents is less than the threshold, the collective opinions can reach a consensus in the social network. Otherwise, the collective opinions will be split into at least two opinion clusters. Secondly, the smaller is the difference of initial opinions, the bigger the probability of reaching consensus. Last but not least, there is a linear correlation between the number of the final opinion clusters and the size of a social network. The linearly increasing rate equals approximately the fraction of the close-minded agents.3. A dominant-submissive heterogeneous bounded confidence model is built. From the social classification theory, the social network is divided into dominant agents(opinion leaders) and submissive agents(ordinary followers) according to the hierarchy of the society. The studies show that the existence of the opinion leaders will influence remarkably the evolution process and the final state of the collective opinions over the social network. Secondly, the guiding power of the opinion leaders mainly depends on the social agents with smaller confidence levels and the confidence degree of the submissive agents on the opinion leaders in a heterogeneous society. Finally, the size of the social network and the fraction of the opinion leaders have no evident influence on the guiding power of the opinion leaders.4. An opinion dynamics with environmental noise is built for the opinion evolution under emergency. The part takes the key feature of emergencies, i.e., uncertainty, as a variable to describe the external environment change, and analyzes the influence of the environmental noise on the collective opinion evolution. Firstly, the probability of reaching a consensus is analyzed for the collective opinions under emergencies. As the external environment uncertainties increase, the social agents form a herding behavior distinctly. Secondly, the guiding power of the opinion leaders is not stable with the influence of the external environmental noise. When the environmental uncertainty is large enough, the guiding power of the opinion leaders does not increase and tends to decrease instead. Finally, when some other conditions are fixed, the relationship between the size of the social network and the evolution of the collective opinions seems to have a comparative stability under the influence of the environmental noises.