Disaster Response Autonomus Modeling
This project develops a hybrid NLP framework for understanding how crisis-related information evolves across social media over time. Using the HumAID dataset, BERTopic identifies emerging topics and tracks their temporal patterns, while LDA provides complementary topic modeling for interpretation. RoBERTa-based classification further analyzes crisis-related content, enabling the study of topic emergence, peak activity, persistence, and information diffusion. The framework provides an analytical approach for extracting actionable insights from social-media data to support crisis monitoring, disaster response, and emergency management.
More work
