A Classification of Disaster Responses Based on an Analysis of Data from Social Media

Abstract

During any natural disaster, a lot of information is created on social media; users produce a lot of information, such as Twitter, to post textual and multimedia content to report updates about injured or dead/missing people, needs and other information types. However, in the past, research in this field has not had much data available; for this purpose, we will require entirely different approaches, tools, and techniques to help inform decision-making under uncertain conditions. The fundamental target of our research in this field is to improve disaster relief efficiency and attention and extract useful information from social media data, like public attitude toward disaster response and the public demands for the targeted based on properties such as needs, damages, etc. In this study, public perception is assessed qualitatively by manually classifying, which contains lots of information like demand for target relief supplies, satisfaction with the disaster response, and public fairness. So, by using public tweets, are analyzed using different machine-learning models. And to better provide the decision maker with the appropriate model, the comparison of different machine learning models based on computational time and prediction accuracy is conducted.