Disaster Management From Social Media Using Machine Learning

In recent years natural calamities and disasters resulted in tragic loss of mankind and caused unpredictable infrastructural damages. The process of identifying damage and casualties allows first responders to efficiently allocate resources and save as many lives as possible. The success of a disaster management process is largely dependent on getting the accurate and timely information about the disaster status, the affected people and the environment in the surrounding areas. Social media is one of the ways of communication in the course of catastrophes. During such catastrophe, bystanders and affected people post instant updates including death toll or reports of injured people, infrastructure damage, requests for immediate necessities like food, water, shelter, donation offers and so on. Social media images are a valuable source of relevant information. However, the huge number of posts and presence of irrelevant posts makes it impossible for emergency responders to manually mine posts for credible information. Therefore, automating the process of information extraction from the posts which are already collected from social media is essential to fully take advantage of this abundance of data. This paper mainly discusses about denoising, and classifying imagery content from social media platforms to help humanitarian organizations in gaining situational awareness, updates and launching relief operations.