SEPM: rapid seism emergency information processing based on social media

With the development of network communication technology and the popularity of social media tools, earthquake-related information has been easily published and disseminated in social networks. This study focuses on obtaining this information and providing guidance for earthquake emergency work. A processing model is proposed to obtain earthquake information from social networks. First, a configuration-driven data acquisition module is designed to acquire earthquake information. Second, according to the characteristics of earthquake information in social media, a seismic emergency thesaurus is selected, and weight is calculated. To solve the low accuracy of inter-class classification, an improved mutual term frequency–inverse document frequency (MTF–IDF) algorithm is proposed. Finally, the thesaurus database is used to classify the acquired earthquake information. By taking the Lushan and Jiuzhaigou earthquakes as examples, the improved MTF–IDF algorithm shows a better effect on the selection of seismic keywords than the traditional TF–IDF algorithm; the F1-measure in classification has increased from 79.86 to 86.93%. The proposed model can rapidly and easily acquire and classify earthquake information according to different sources, which can provide timely information and support for disaster relief.