Abstract:

Once an emergency network public opinion crisis breaks out， it will cause unrest among the people and hinder social stability. In order to forecast crisis contacts in a timely and effective manner， prevent the outbreak of public opinion， and pre-control the escalation of the situation， it is urgent to study crisis warning. Firstly， the early warning block indicators are established based on the network public opinion information attributes of emergencies in this paper， the weights are determined by ANP， and the three-level risk indicators which is from high to low risk （level I， II， and III） are obtained through the cumulative weight ratio in the ABC classification method thoughts. Secondly， the “cascade-parallel” early warning mechanism based on the framework of the early warning system is designed to monitor real-time dynamics and predict the future situation. Finally， taking the “new outbreak in Beijing” as an example， the early warning mechanism is simulated and analyzed by using the isomorphism of the Stochastic Petri Net and the Markov chain. The results show that the nonlinear relationship is found between the real-time value of level I and level II risk indicators and the probability of occurrence of public opinion crisis， as well as the growth rate of risk indicators， and the interactive relationship among the growth rate of level I， level II and level III has produced the critical surface of early warning. According to these analyses， the three-dimensional graph is drawn and the early warning start rule is formulated. In the rule， the interactive relationship among different indicators is considered， the early warning activation level from many aspects is judged， and the alert level is divided into three categories from small to large： yellow， orange and red， so as to provide decision-making reference for relevant departments to take targeted preventive measures in advance.

Key words: ANP, Stochastic Petri Net, network public opinion, crisis warning, Markov chain