OCA - Ocurrences Citizen Assistant

Problem Statement: In emergencies like medical crises, natural disasters, or urgent incidents, people often struggle to find accurate and timely contact information for relevant government authorities. Traditional search methods are slow, fragmented, and not user-friendly during moments of crisis. Solution: Our project offers an AI-powered emergency response system that provides real-time, accurate, and official government contact information through a user-friendly interface. Using state-of-the-art Retrieval-Augmented Generation (RAG), natural language processing, and web crawling, the system empowers users to connect with the right resources in their most critical moments. Core Features: User-Friendly Input: Inputs can be provided as text or audio via a sleek Streamlit-based interface. Accessibility features ensure usability for people with varying technical skills and physical conditions. Advanced Query Understanding: Leverages LLM-powered language models to understand user queries and context effectively. Automatically detects location and matches it with the user's query for precise results. Real-Time Web Crawling: A custom crawler engine retrieves the latest and most accurate government resources, including phone numbers, addresses, and official websites. Intelligent Retrieval and Summarization: Employs a Retrieval-Augmented Generation (RAG) pipeline to extract, summarize, and format information as actionable insights. Generates human-readable responses like "Final Answers," summarizing critical details for users. Streamlined Output Display: Outputs are presented in a clean, visually appealing Streamlit dashboard, highlighting the most relevant "Final Answers" in Markdown format. Supports saving historical results to files for traceability and post-event analysis. Multi-Device Compatibility: Accessible via desktop, tablet, or mobile devices for quick access during emergencies.