PDF-QuestionAnswer

A Python-based PDF Question-Answering system powered by Retrieval-Augmented Generation (RAG).

This project extracts text from PDFs, chunks it, generates embeddings (using Hugging Face or OpenAI models), stores them in Qdrant (an open-source vector database), and performs efficient similarity search to retrieve relevant context. The retrieved context is then used to augment prompts for a Large Language Model, delivering accurate and document-grounded answers.

Perfect for building custom document-specific QA applications.

Features

• PDF text extraction and intelligent chunking
• Support for Hugging Face or OpenAI embeddings
• Vector storage and fast similarity search with Qdrant
• Full RAG pipeline for reliable question answering
• Interactive Streamlit web interface

Tech Stack

• Python
• Qdrant (vector database)
• LangChain (RAG orchestration)
• Streamlit (UI)
• Hugging Face Transformers / OpenAI API
• PyPDF2 or similar for PDF processing

Installation

1. Clone the repository:

   git clone https://github.com/1rishu0/PDF-QuestionAnswer.git
   cd PDF-QuestionAnswer
   

2. Create a virtual environment:

   python -m venv venv
   source venv/bin/activate  # On Windows: venv\Scripts\activate
   

3. Install dependencies:

   pip install -r requirements.txt
   

4. Set up environment variables (required for OpenAI usage): Create a .env file in the root directory:

   OPENAI_API_KEY=your_openai_api_key_here
   
   

5. Run Qdrant using Docker:

   docker run -p 6333:6333 -v $(pwd)/qdrant_storage:/qdrant/storage qdrant/qdrant
   

Demo Screenshots

The Streamlit interface for uploading PDFs and asking questions.

Usage

Streamlit Web App

Launch the Interactive Interface:

streamlit run streamlit_app.py

Steps:

• Upload your PDF document
• Wait for the document to be processed and indexed
• Start asking questions about the content

Command-Line Interface

Run the CLI version:

python main.py --pdf_pdf path/to/your/document.pdf

Steps:

• The PDF will be loaded and indexed
• Follow the console prompts to ask questions

Configuration Tips

• For local/free usage: The project supports Hugging Face embedding models (no API key needed)
• For better performance: Use OpenAI embeddings/LLM by setting OPENAI_API_KEY
• Persistent Qdrant storage: The Docker command above mounts a local folder for data persistence

Project Structure

.
├── streamlit_app.py      # Streamlit web interface
├── main.py               # CLI entry point
├── data_loader.py        # PDF loading, text extraction & chunking
├── vector_db.py          # Qdrant integration (indexing & retrieval)
├── custom_types.py       # Custom type definitions
├── pyproject.toml        # Dependencies (Poetry)
└── README.md             # This file

Contributing

Contributions are welcome! Feel free to:

• Open issues for bugs or feature requests
• Fork the repo and submit pull requests
• Improve documentation or add examples

License

This project is licensed under the GPL-3.0 License - see the LICENSE file for details.

You can directly copy and paste this content into your repository's `README.md` file. All sections after **Installation** are now fully formatted with proper Markdown headings, lists, code blocks, and bold text for better readability.