Crisis and Humanitarian Computing

During time-critical events such as natural disasters, situational awareness becomes a crucial task for authorities to understand the situation as it unfolds. Relying on traditional information sources could potentially delay gathering critical information such as infrastructure damage, info about injured or deceased people, urgent needs of affected people, etc. This delay can lead to a much more significant loss in terms of human lives and the economy.

With the proliferation of social media platforms such as Facebook, Twitter, Instagram, and many others (~800 others), bystanders and eyewitnesses in and around crisis events actively post situational information. Research studies have shown that this online information could be beneficial for humanitarian organizations to enhance situational awareness. For instance, Twitter's 280-character message is a fantastic service that proved it to be anything to many events, disasters, people, celebrities, brands, etc. It was the Twitter that triggered Arab upheaval, saved many lives during the Sandy hurricane in the US, and a devastating earthquake in Nepal in 2015.

Although humanitarian organizations would like to use social media into their disaster response efforts, processing and extracting useful information from social media streams is still a challenging task for several reasons such as information overload, veracity, and variety. Both textual messages and images shared on social media could be useful to gain situational awareness as well as to extract actionable information. We believe that if analyzed timely and effectively, these small bits of information can collectively help humanitarian organizations in their decision-making processes and eventually can save hundreds of lives.

With this motivation in mind, my research focuses on developing computational methods, techniques, and technologies to process big crisis data on social media. I focus on developing novel text classification, data mining, machine learning, and deep neural networks techniques, which can help stakeholders to gain situational awareness and actionable information to improve their decision-making processes.