Earthquake Disaster Data Mining And Application Of Rapid Intensity Assessment Based On Social Media

Earthquake disaster is one of the most serious disasters.China is one of the countries which suffered the most serious earthquake disasters in the world.Earthquakes in China distribute widely,occur frequently and cause serious losses,and these facts are the basic national conditions of our country.Thus,reducing the loss of earthquake disasters is an important reality that we must face during economic development.After destructive earthquakes,obtaining disaster information efficiently and quickly,and evaluating the intensity are the the key problems of earthquake emergency.Exploring these problems have always been an important research topic in academia,as well as the most concerned topics of governments at all levels.In recent years,with the rapid development of mobile Internet technology,social media platform with massive data provides a new perspective and an important way for earthquake disaster information acquisition and earthquake intensity rapid assessment.Because of the characteristics of social media data such as mass,space-time,interaction,strong diffusion and integration,the public can freely express their views,opinions and feelings through social media,which has virtually accelerated the sharing and dissemination of disaster information.Mining the massive data of social media spontaneously contributed by users after the earthquake will bring the thought of crowd sourcing into earthquake emergency response,finally becomes an important way to effectively enhance the ability of earthquake disaster acquisition.Therefore,on the basis of absorbing the ideas and methods of information science,engineering seismology,management science and engineering,and statistics,this paper makes a thorough study on the problems of data capture,discriminating and storage,space-time characteristics and thematic distribution characteristics of post-earthquake disaster situation in social media,furthermore,it combines machine learning and proposes a rapid seismic intensity assessment method based on social media data.The main objective of this study is to analyze the characteristics and regularities of earthquake disaster data on social media platform in mainland China,so as to promote the development of social media earthquake disaster data mining as a new emerging research field,and explore a new rapid assessment of intensity based on social media data.This method is expected to improve the efficiency of earthquake emergency rescue work and provide reference for earthquake emergency command and decision making.On the basis of fully absorbing and summarizing the previous achievements,thispaper takes the destructive earthquake disaster data of Weibo as the research object,and tries to explore and solve the key scientific issues of earthquake emergency and loss assessment.The main research work and innovative achievements are as follows:1.A multi-strategy social media disaster data acquisition scheme is proposed.The first earthquake disaster database and management platform of social media in mainland China is established based on Sina Weibo mobile terminal,which lays an important foundation for the research work in this field.Taking Sina Weibo,the largest and most popular social media platform in China as an example,this paper summarizes and analyses the data acquisition methods of Weibo.After that,a multi-strategy social media earthquake disaster data acquisition method combining microblog commercial API,web crawler,intensity attenuation relationship and historical earthquake intensity distribution vector map is proposed.We have captured the earthquake-related microblog data within 72 hours after 206 destructive earthquakes in China since 2010,established the first earthquake social media disaster database and management platform in mainland China,and realized data visualization.At the same time,the vector maps of 26 destructive earthquakes intensity distribution in the mainland of China are collected and sorted out.On this basis,the location microblog extraction and map matching are completed.The database established in this paper contains text content data and user relationship data.It is rich in content,detailed in information and easy to download and use.The establishment of the database provides valuable basic information for future earthquake disaster acquisition and data mining work.2.Based on social media data,this paper analyzed the temporal characteristics,spatial distribution characteristics,temporal and spatial variation characteristics and thematic characteristics of the disasters caused by destructive earthquakes in China since 2010,and obtained the statistical characteristics and distribution rules of the disasters.Based on the 72-hour post-earthquake microblog data acquired in Chapter 3,the descriptive mining is carried out.It also did the time and space analysis of the overall data while the spatial and temporal characteristics of the location microblog data are analyzed based on thermodynamic diagram,and the subject clustering analysis is made by using K-means method,and the distribution of different disaster themes is mastered.The statistical characteristics and rules of social media disaster data of destructive earthquakes in China in recent years are obtained.3.Based on the social media data and the artificial neural network algorithm in machine learning,a fast seismic intensity evaluation model is established,and adata-driven seismic intensity evaluation method is proposed.The idea of multi-classification problem and text mining method in machine learning are introduced into the rapid seismic intensity assessment.The overall framework and process of the rapid seismic intensity assessment method based on post-earthquake social media data are proposed.Using 20 destructive earthquakes of Sina Weibo data from 2010 to now as samples,the eigenvector matrix of Weibo text data is constructed,and the corresponding relationship between data and intensity zoning is established.The fragmented and semi-structured Weibo text data is transformed into space vector form which can be used as input for classification problems,thus forming machine science.Based on the artificial neural network(ANN)algorithm,a fast seismic intensity evaluation model is trained from the structured data set.The accuracy of this model can reach 81% by performance test and 67% by empirical analysis.This data-driven seismic intensity rapid assessment method proposed in this paper can meet the actual needs of earthquake emergency rescue relatively well in terms of timeliness and accuracy.The academic contribution and application value of this thesis mainly includes opening up a new way to acquire information of earthquake disaster,providing a new research idea for rapid acquisition of disaster and rapid assessment of earthquake intensity in earthquake emergency response,and putting forward a new method for rapid assessment of earthquake intensity.The results have important application value in both earthquake emergency and governmental management.