Enabling Rapid Disaster Response Using Artificial Intelligence and Social Media

Disasters and emergencies often bring unanticipated situations for the members of public and formal response organizations. During such times, access to rapidly changing information plays an important role to understand the situation as it unfolds. However, information scarcity, especially in the early hours, hinders this task and ultimately delays response operations. Social media platforms such as Twitter and Facebook are gaining attention of both formal response organizations and affected individuals. People use social media to share a variety of information online including requests for help, and other urgent needs such as food, water, and shelter. This online information can be useful for humanitarian organizations if processed timely. This work presents a number of Artificial Intelligence (AI) technologies developed at the Qatar Computing Research Institute (QCRI) to aid disaster response and management. These technologies include machine learning platforms to automatically process textual messages (e.g., tweets) into humanitarian categories such as reports of needs, injured or dead people, and to process imagery data. The platforms work in real-time to collect and analyze data on social media to help humanitarian organization gain situational awareness and extract actionable information. These technologies were deployed during the Hurricane Maria and in this paper we present our findings and insights from this real-world natural disaster.