Research On Mining And Responsing To Dynamic Network Public Opinion Of Emergencies

China is now facing dramatic system transition and social transformation. As a criti cal period of economic development and social contradictions, all kinds of unexpected e vents occur. Therefore, scientific and effective emergency management is needed to gua rantee the healthy development of the society.The twenty-first Century is the network era of rapid development information technology and information resources, the occurrence of emergencies must be accompanied by a lot of network public opinions.Under the network environment, the network public opinion with different carrier types and wide sources, presents a features of dynamic and structural complexity. Network public opinion makes the evolution mechanism of emergency more complex, focusing or expansion, the formation of secondary, derivative event, become the focus and difficulty of emergency management. According to the characteristics and the evolution rules of network public opinion, this paper mainly mined and analyzed two aspects of content and structure of network public opinion for emergencies about how to make the network public opinion more targeted and scientific, focusing on the following questions: First of all, what are the factors of emergency network public opinion? What are the similarities and differences of the network public opinion attention heat about the different emergencies? Then, what are the features of the network public opinion node network, and how to identify the key node? How does the key node classify and evolute? Finally, based on the above research of data mining and case analysis of the evolution, how to deal with the emergency management department on the network public opinion?Network public opinion research is a hot topic for emergency management today. Scientific research methods should be adopted because of the complexity and rapid development of the network environment. So,this paper combined with information dissemination, management science, computer science, behavioral science, psychology research in order to obtain the pattern and scientific response theory during the development of the emergency network public opinion, The main contents are as follows:First,In the omni media network environment, the network public opinion is dynamic and complexity. The occurrence, development and disappearance of public opinion showed some regular evolution pattern. Therefore, to deal with the public opinion on the network, we must start from the information source, which is the original data of network public opinion, construct the theoretical response model of network public opinion, including the source layer, mining layer and response layer, so that the model is scientific and feasible.Second, Constructed network public opinion heat system containing the emergency response level from multiple factor(emergency factor, network media factor, the netizen factor, government guiding factor), using GooSeeker, DataScraper and MetaStudio methods for data acquisition about 8 events of different types and different response levels. Results showed: for different emergencies, the network public opinion heat contribution by factors is unbalanced; By introducing the influence factor and the focus factor in the formula, the event function types is summarized as three types of attenuation type, continuous type and the complex type. The relevance between the emergency response level and network public opinion heat are consistent and inconsistent from the results, and the causes are analyzed.Third, In order to carry on network structure analysis, key nodes identification and classification of network public opinion, this paper brought up the technical method which integrating data mining module, data structure module, node measuring and recognition module, key nodes influence calculation module in forum, key node classification module etc., involving the algorithm and software such as GooSeeker, Gephi, LeaderRank, Notepad++. Taking The 7·23 train accident as the example for detailed analysis, revealed the complexity of structure, scale-free and community structure etc. of network public opinion. By the classification to the key nodes combined with the key node influence in forum, this paper got two kinds of key nodes as "web celebrities" and "event focus".Combined with above-mentioned analysis, the paper put forward the concept, content and characteristics of the network public opinion response information flow; Put the dynamic classification method of network public opinion heat as the effective supplement to the emergency classification method; Because the network media agenda have the transmission advantage by setting structural holes, we can guide the direction and trend of the network public opinion scientificly through the network media agenda setting and key nodes guiding. The paper constructed the dynamic multi governance model of network public opinion with the promotion strategy of the main factors.