Urocure - Urology Chatbot

Urocure is a specialized chatbot designed for the field of urology. This Python project utilizes a Streamlit web app, allowing users to interact with the AI and ask questions related to urology. The chatbot is powered by RAG (Retrieval-Augmented Generation) and has been trained on a Chinese dataset containing over 2000 entries. The dataset is stored in MongoDB Atlas as a vector database.

Features

• Streamlit Web App: User-friendly interface for interacting with the urology chatbot.
• RAG Model: Utilizes Retrieval-Augmented Generation for improved responses.
• Chinese Dataset: Trained on a diverse dataset of 2000+ entries in Chinese. Dataset link: RJUA QADatasets.
• MongoDB Atlas: Stores the dataset as a vector database for efficient retrieval.
• Cloudflare Translation API: Translates English requests to Chinese and vice versa.
• CloudFlare Text Generation API: Generates Text.

How It Works

1. User Prompt: Users send their queries in English through the web app.
2. Translation: Cloudflare Translation API is used to translate the English request into Chinese.
3. RAG Embeddings: The translated request is passed through the RAG model to generate embeddings.
4. Embedding Conversion: Chinese embeddings are converted back to English based on prompt engineering.
5. MongoDB Atlas Integration: Utilizes MongoDB Atlas as a vector database for storing and retrieving the Chinese dataset.
6. Streamlit Interface: The user interacts with the chatbot through a user-friendly Streamlit web app.

Setup

1. Clone the repository:

2. Install dependencies:

pip install -r requirements.txt

3. Set up MongoDB Atlas:

   • Create an account on MongoDB Atlas.
   • Create a new cluster and obtain the connection string.
   • Update the connection string in config.py.

4. Set up Cloudflare Translation API:

   • Obtain an API key for Cloudflare Translation API.
   • Set the CLOUDFLARE_API_KEY environment variable.

5. Run the Streamlit web app:

streamlit run app.py

Visit http://localhost:8501 to interact with Urocure.

Configuration

• ATLAS_CONNECTION_STRING: Set the MongoDB Atlas connection string in config.py.
• OPENAI_API_KEY: Set the OpenAI API key in .env file.
• CLOUDFLARE_API_KEY: Set the Cloudflare API key in .env file.

License

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

Feel free to contribute and enhance the capabilities of Urocure!