| The unexpected public crisis takes risk and difficulties to the crisis management, because of its characteristics such as explosion, particularity, environmental complexity, evolving uncertainty and group diffusion. From the perspective of online public opinion, this research focuses on the mechanism between unexpected public crisis and online public opinion. It provides theoretic supports for the crisis management. The main results of this study are as follows:(1) Putting forward a dynamic mechanism of online public opinion evolution in which internal driving power and exogenous driving power are the main part.Through the analysis of mechanism and the evolution path of unexpected public crisis and online public opinion, from the theoretical perspective, we put forward a dynamic mechanism of online public opinion evolution which mainly includes internal driving power and exogenous driving power. The power which impel the situation of online public opinion begin to evolution including the destructive power from the crisis itself, the impetus from the network by netizen and network media, and the regulation power from the government agencies; According to their essential characteristics, we divide these three elements into two parts, the internal driving power and exogenous driving power. At the beginning, we deconstruct34elements for internal driving power and exogenous driving power by communicating with28experts. Through statistic analysis and summarizing, we choose22elements from34to be the most important ones.The main elements of internal driving power include the sensitivity and publicity of the crisis.In the experimental research for sensitivity factors, we studied60unexpected and influencial public crises from2008to2012, and concluded that there were12most sensitive factors; The mainly constituent elements of exogenous driving power rank into three first-class indicators, six second-class indicators, four third-class indicators and five forth-class indicators. This result provides the theoretical supports for the other conclusions in this research.(2) We established a dynamic model based on the mechanism between unexpected public crisis and online public opinion.After studying the complicated relationship between unexpected public crisis and online public opinion, we built a dynamic model by system dynamics method describing the mechanism between unexpected public crisis and online public opinion from the macro perspective. This system dynamics model is constructed by three subsystems, including the government agencies, network media and netizen. There are35variables,39feedback loops and one kind of typical systems archetype in the SD model. The data come from36unexpected public crises and85questionnaires (reliability coefficient a>0.7, P<0.05). In order to validate the reliability of the SD model, we compare the true value and the predictive value of two variables, which are the number of BBS boots and network news. The MAPE value is between10%and20%, proving the reliability of SD model in the previous literature, there was no similar research. This study is definitely positive and innovative. We analyze four moderator variables from12variables, and get16conclusions. These four variables are the sensitivity of the crisis, the publicity of the crisis, the credibility of the government, and the query degree of for the crisis. This research concludes seven suggestions for dealing with unexpected public crisis and directing the online public opinion.(3) We built an early-warning model to predict the hot degree of online public opinion on the basis of Bayesian network.After considering the features of Bayesian network, such as the expressing ability for variables’complex relationship, the calculating ability for the uncertainty probability and the reasoning ability for cause-effect relationship, we built the early-warning model to pridict the hot degree of online public opinion on the basis of Bayesian network from the medium perspective. There are three steps. Firstly, we construct the structure of the Bayesian network. As we all know, the internal driving power and exogenous driving power push crisis into a hot online topic. As a result, we divide Bayesian network into three layers. They are results layer (the results of early-warning), the structural layer (the main structure of Bayesian network) and the data layer (the simulation data for the medal). And every data in Bayesian network is ranked into low, middle or high level. Secondly, we determine the conditional probability of the Bayesian network. The data for conditional probability study come from83unexpected public crises. Converging experts’experience andvariables’features, we confirm the standard to discretize the continuous data. After comparing many kinds of algorithms, finally we use EM algorithms for conditional probabilitystudy. Thirdly, using Netica for simulation. We use the testing set to verify the effect of the model and we ultimately prove the reliability of our model. In the simulation, we supposed that we gain the new evidences (the sensitivity of the crisis, the reliability of the crisis and the unselectable of the crisis), and the simulation result show that the probable situation of the online public opinion is low (24%), middle (30.8%) and high (45.2%). So the conclusion is that the situation of the online public opinion will be hot. As stated previously, we put forward a new method for online public opinion early-warning study. This method will make the early-warning process more intelligent, automatic and efficient.(4) We built a coupling model for unexpected public crisis and online public opinion based on complex systems theory.From the micro perspective, we built the coupling model for unexpected public crisis and online public opinion based on complex systems theory. There are three steps. Firstly, we contribute an index system to quantize the degree of the internal driving power and exogenous driving power. And we use BP neural network algorithm for calculating the weight of the indexes. Secondly, we establish the coupling model. Based on the index system, we established the coupling model to quantize the coupling degree between internal driving module and exogenous driving module. We select16unexpected public crises to calculate their coupling degrees. Finally, we found that there is strong or weak coupling relationship among16crisis’s modules:6crises have the highly coupled relationship, and one crisis has the extreme coupling relationship. Thirdly, we make the empirical research. In the empirical research, we choose two quantitative indicators including the number of network news and the number of network news reply, which can reflect the social influence of the unexpected public crisis. And then, we use SPSS to analyze the correlativity between the coupling degree of the modules from the16crises and the two quantitative indicators. Ultimately, we found there is a significantly positive correlation among them. This result proves that the greater the coupling degree is, the higher the social influence of the unexpected public crisis is. Based on these, in order to reduce the negative social influence to the lowest level, we give four applicative suggestions to cut off the coupling, They are establishing the sensitive factors thesaurus, building the early warning system, regulating effectively by government and the setting up high-efficiency online public opinion monitoring platform. |
PDF Full Text Request