Area Definition and Public Opinion Research of Natural Disaster Based on Micro-blog Data

Natural disasters have long disturbed people’s normal lives. How to effectively identify and assess natural disaster-affected areas and further control network grievances is of great significance. This paper takes the floods in Hunan Province of China in 2017 as an example, based on the data sources of Sina Weibo, and combines text analysis and image analysis technology to define the disaster scope of natural disasters, and then uses sentimental analysis method and correlation analysis method to study the network public opinion in the disaster area. In this paper, the average accuracy and recall rate of the model which combines image analysis and text analysis to define the disaster area are 72.10% and 81.59% respectively, which effectively realizes the definition of flood disaster area in Hunan Province. In addition, this paper finds out the law of the development and change of network public opinion, and puts forward the opinions and suggestions on the determination of disaster area, the relief of disaster and the control of network public opinion.