Disaster Response Planning with LLM-Powered Knowledge Synthesis and Rumor Detection

This thesis project focuses on developing a comprehensive disaster response system that leverages Large Language Models for intelligent decision-making during crisis situations.

The system integrates local LLMs with real-time web data collection to provide accurate, up-to-date information while filtering out misinformation and rumors that commonly spread during disasters.

Built using Streamlit for an intuitive interface, the application employs Chain-of-Thought reasoning and RAG (Retrieval-Augmented Generation) to synthesize knowledge from multiple sources and provide structured disaster response recommendations.