Context-Aware Geospatial Data Retrieval using LLMs

A sophisticated web application that leverages Large Language Models (LLMs) to provide intelligent, context-aware geospatial data retrieval and visualization. This project combines natural language processing with interactive mapping to create an intuitive interface for location-based queries.

Features

• Natural Language Processing: Interact with the system using everyday language to query location information
• Context-Aware Responses: Maintains conversation history for more relevant and contextual responses
• Real-time Geocoding: Converts location names to precise coordinates using Google Maps API
• Interactive Map Visualization: Dynamic map interface that updates based on location queries
• Real-time Data Retrieval: Access current information about locations using SerpAPI
• Responsive UI: Modern, user-friendly interface with collapsible sidebar

Output Samples

Sample 1: Location Query and Response

Query about restaurants with map visualization

Sample 2: Multiple Location Display

Displaying multiple locations with detailed information

Sample 3: Interactive Map Features

Interactive map with location markers and information

Sample 4: Context-Aware Response

Context-aware response with location details

Sample 5: Real-time Data Integration

Real-time data integration with map visualization

Tech Stack

Backend

• Python 3.x
• Flask (Web Framework)
• OpenAI GPT-4 (LLM Integration)
• Google Maps Geocoding API
• SerpAPI (Real-time Data)
• Agno Framework (LLM Agent Management)

Frontend

• React
• TypeScript
• Tailwind CSS
• Interactive Map Component
• Vite (Build Tool)

Getting Started

Prerequisites

• Python 3.x
• Node.js and npm
• API Keys for:
  • OpenAI
  • Google Maps
  • SerpAPI

Installation

1. Clone the repository:

git clone https://github.com/yourusername/geospatial-llm.git
cd geospatial-llm

2. Backend Setup:

cd backend
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate
pip install -r requirements.txt

3. Frontend Setup:

cd frontend
npm install

4. Environment Variables: Create a .env file in the backend directory with:

OPENAI_API_KEY=your_openai_api_key
GOOGLE_MAPS_API_KEY=your_google_maps_api_key
SERPAPI_KEY=your_serpapi_key

Running the Application

1. Start the Backend:

cd backend
python server.py

2. Start the Frontend:

cd frontend
npm run dev

The application will be available at http://localhost:5173

Usage

1. Open the application in your web browser
2. Type your location-based query in the chat interface
3. The system will:
   • Process your query using the LLM
   • Extract relevant locations
   • Convert locations to coordinates
   • Display the locations on the interactive map
   • Provide detailed information about the locations

Example queries:

• "Show me the best restaurants in Pune"
• "What are the popular tourist spots in Mumbai?"
• "Find hospitals near Koregaon Park"
• "Plan 3 days trip to Goa."

Configuration

The system can be configured through various environment variables and settings:

• num_history_responses: Number of previous responses to maintain for context (default: 4)
• Map default center and zoom level can be adjusted in App.tsx
• LLM model parameters can be modified in agent.py

License

This project is licensed under the MIT License - see the LICENSE file for details.

Citation

If you use this project in your work, please cite it as follows:

@software{geospatial_llm_2024,
  author = {Aditya Shingote, Shishir Narkhede, Sunil Kumawat, Kaushal Borkar,},
  title = {Context-Aware Geospatial Data Retrieval using LLMs},
  year = {2024},
  url = {https://github.com/Sunilk240/geospatial-llm}
}

Or simply mention us in your acknowledgments:

This project uses the Context-Aware Geospatial Data Retrieval system developed by Aditya Shingote, Shishir Narkhede, Sunil Kumawat, and Kaushal Borkar.

Authors

• Aditya Shingote
• Shishir Narkhede
• Sunil Kumawat
• Kaushal Borkar

Acknowledgments

• OpenAI for GPT-4 API
• Google Maps Platform
• SerpAPI
• Agno Framework
• React and TypeScript communities