Twitter-Based Disaster Response Using Machine Learning

Abstract

Twitter, a microblogging platform, receives real-time information via informal conversations, and it has, accordingly, become the main source of data for research studies based on emergency situational awareness. Millions of tweets are posted on Twitter every day, and during disasters, the frequency of tweets relating to an on-going crisis event grows exponentially. This unprecedented increase in the number of tweets during disasters needs to be monitored, identified, processed, and analyzed so that necessary measures can be taken at the earliest to reduce the loss or damage during emergencies. However, due to large voluminous data being available during crisis hours, it is almost impossible for a human to perform these tasks in real time. In this regard, a semi-automated AI-based disaster response system for Twitter data is proposed. The proposed disaster response system would be capable of extracting essential situational awareness information related to a disaster and would also be capable of sketching tentative area of critically affected population.