RagBot with Gemini API

Overview

RagBot is an AI-powered chatbot that integrates Retrieval-Augmented Generation (RAG) with Google's Gemini API. It retrieves relevant documents from a vector database and enhances responses using a generative AI model. This project is built using Streamlit for the frontend and ClickHouse for vector storage.

Live Demo

https://ragbot-with-gemini-api.onrender.com/ - might not work yet

Features

• Extracts and processes text from PDFs and images
• Converts text into vector embeddings for efficient search
• Retrieves relevant documents based on user queries
• Generates AI-powered responses using the Gemini API
• Built with Streamlit for an interactive UI

Technologies Used

• Python (Streamlit, Pandas, PyPDF2)
• Google Gemini API for text generation
• ClickHouse for vector storage and retrieval
• Google Generative AI for content embedding
• PyPDF2 for PDF text extraction
• OpenCV & Tesseract OCR for image processing

Installation

1. Clone the repository:

   git clone https://github.com/Vinodhariharan/RagBot-with-Gemini-API.git

2. Navigate to the project directory:

   cd RagBot-with-Gemini-API

3. Install dependencies:

   pip install -r requirements.txt

4. Set up environment variables:
   • GEMINI_API_KEY: Your API key for the Gemini API
   • CLICKHOUSE_HOST, CLICKHOUSE_USERNAME, CLICKHOUSE_PASSWORD: ClickHouse database credentials

Usage

1. Run the Streamlit application:

   streamlit run app.py

2. Upload a PDF or an image to extract and store content in the vector database.
3. Enter a query in the chatbot interface to retrieve relevant documents and get AI-generated responses.

Project Structure

RagBot-with-Gemini-API/
│── app.py                    # Main Streamlit application
│── requirements.txt          # Python dependencies
│── utils/
│   │── vector_storage.py      # Vector DB operations
│   │── pdf_processing.py      # PDF text extraction
│   │── image_processing.py    # Image text extraction
│── models/
│   │── gemini_integration.py  # Gemini API integration
│   │── rag_retrieval.py       # Retrieval logic
│── pages/
│   │── chat.py                # Chat interface
│   │── upload.py              # Upload interface

Contributing

Contributions are welcome! Feel free to fork the repository and submit pull requests.

License

This project is licensed under the MIT License.