Research On Method Of Network Public Opinion Monitoring Of Emergency Based On Micro-blog

Emergencies occur frequently result in tremendous social loss, therefore more and more attentions have been paid by people. Especially with the rapid development of Internet, network has become the important platform for public to express willingness, participate in economic and social political life, and to carry on the supervision of public opinion. Via Internet, public are increasingly involved in public emergency, and played an cannot be neglected role on promoting the development and changes of events. So the monitoring of network public opinion triggered by emergency has become the urgent need of emergency management.However, the existing monitoring research of emergency network public opinion has achieved certain results but there are still many problems. First of all, existing research always describe the evolution process and situation of public opinion from the perspective of qualitative, and there is no quantitative judgment method about the each phase of network public opinion. Especially for the problem of how to judge the possible outbreak time and palces of public opinion, there is still no effective solution. Second, the existing monitoring method based on index system is still in the stage of theoretical research and their practical application value is not high. Due to the contents defined by indicators are relative broadly, it is always hard to get many indicators data in actual practice. Again, existing public opinion monitoring methods and monitoring system which based on data mining and natural language processing, the analysis process of them are only based on simple mathematical statistics method. So, when they obtained the public opinion data, it is always lack of effective mathematical model to further analysis and judgment of the obtained data.Therefore, for the shortcomings of the existing research, this article will take micro-blog as the data source, by introducing the scan statistic method for emergency network public opinion to judge the possible outbreak time and outbreak spatial areas of the evolution process of emergency network public opinion. First, the montoring factors of micro-blog were analyzed and the micro-blog item information and user information with characteristic information of emergency were extracted, taking these as the foundations of how to judge the information size of public opinion in a particular period of time, so that which can provide basis for the next calculation of monitoring. Second, by introducing a scan statistic method, the number of micro-blogs and users for each of window were scanned real-time. And then by calculating the log likelihood ratio for each of window respectively, to position the cluster time period and cluster spatial areas of public opinion and determine the possible outbreak time and spatial areas of public opinion. Finally, an example is illustrated to verify the effective of the proposed method.This article first applied scan statistic method in the field of monitoring network public opinion of emergency of micro-blog, which provided a new theoretical method for public opinion monitoring field, and it has strong practical significance. The established time and spatial scan statistic calculation model solving the problem that current monitoring study of network public opinion of emergency cannot judge the outbreak time and outbreak spatial of public opinion effectively. The proposed method provided technical supports for the personnel of emergency management to timely warn and effectively deal with the network public opinion of emergency.