Classification of Microblogs for support Emergency Responses

Emergency Response encompasses the decisions and actions taken to deal with the immediate effects of an emergency. How to get onsite information about the disaster quickly has been the key issue affecting the effectiveness of the emergency response. As a new form of social media, microblogging is becoming an applicable source of information for emergency response management. This paper discussed how to use text classification techniques to divert useful microblogs to relevant departments. At first we pre-proceed microblogs in order to fit the classification model, and then the Naive Bayes model was adopted for the classification. Microblogs of Yushu earthquake in China was used as the study case. As the results of our proposed model for the case, we achieved over 85% recall rate in average. Our experiment shows the feasibility of the method proposed and efficiency of the classification.