Social media can generate a wealth of real-time information during crisis events. However, large volumes of data and irrelevant content create practical challenges.
We are working on this problem by building a suite of deep learning models to classify Tweets during extreme weather events. The models help to turn unstructured social media data into actionable insights for crisis managers.
We published a journal article "Using deep learning and social network analysis to understand and manage extreme flooding." This work was featured in a recent press release and featured in various news articles.