FloatChat is an AI-powered conversational interface for exploring ARGO float oceanographic data. It allows users to query, explore, and visualize oceanographic information using natural language.
- Chatbot-style interface with ChatGPT-style bubbles
- Floating input at the bottom for easy conversation
- Interactive visualizations:
- Float locations on a map
- Pressure and other parameter time-series
- Backend:
- PostgreSQL for structured ARGO data
- ChromaDB for vector search and metadata retrieval
- Retrieval-Augmented Generation (RAG) for natural language query handling
- Python 3.x
- Streamlit – Frontend dashboard and chat interface
- Plotly & Leaflet – Interactive plots and maps
- PostgreSQL – Structured ARGO dataset
- ChromaDB – Vector database for semantic search
- Flan-T5 / Local LLM – Query interpretation
Try the following queries in the chat interface:
- Show me the temperature and pressure profiles of ARGO floats near the equator.
- Compare salinity levels in the Arabian Sea and Bay of Bengal over the last 6 months.
- What are the nearest ARGO floats to latitude 10°N and longitude 75°E?
- Show me pressure vs depth for floats in the Indian Ocean.
- Plot the pressure trends of ARGO floats in March 2023.
Activate your virtual environment & D:/my_project/venv/Scripts/Activate.ps1
Run the Streamlit app using cd D:\floatchat then streamlit run front_end/app.py finally Open in your browser: http://localhost:8501/
📌 Notes Demo uses a subset of 500 ARGO floats for speed.
Designed for internal hackathon PoC.
Fully functional chat + inline visualizations for Indian Ocean ARGO dataset.
🔗 Future Work Full ARGO dataset integration
Additional BGC and satellite data
Enhanced RAG pipeline for advanced queries
Improved UI/UX and multi-modal visualizations