Real Time Twitter Based Disaster Response System for Indian Scenarios

Abstract

Twitter is a popular social media platform with more than 1 million daily active users. Mostly, all breaking news is posted earlier in twitter than any mainstream media. Hence, this microblogging social network experiences a deluge of information flow during natural disasters. Situation based mining of information from the twitter data, can play a significant role in disaster response and recovery. The large volume and velocity of data flow on twitter during disaster makes it tedious for the disaster rescue volunteers to manually analyze and retrieve information from them. An automated system that could retrieve relevant information from this enormous twitter data during a disaster, could be useful for the disaster relief volunteers to accomplish their duty efficiently amidst the chaos. During disasters, the volunteer's team may service more efficiently, if they had a classification based on the victims who request for donation and those who request rescue. In this paper, we propose an artificial intelligence based real time disaster response system-Disastro, which assists the volunteers by identifying the relevant tweets from the real time twitter data and classifying them under the domains "rescue" and "donation". Disastro is empirically validated across various machine learning algorithms for classification using the tweets posted during Chennai rains and Kerala floods. The versatility of Disastro across different disasters and its improved classification accuracy makes it flexible and robust to handle any location-based emergencies.