PDF Document Q&A System with RAG and Gemini

Intelligent document analysis powered by RAG and Google's Gemini AI

Features

• 📄 PDF document processing with automatic chunking
• 💡 Natural language question answering
• 🎯 Source citations with page numbers
• 🔍 AI-powered analysis:
  • Sentiment analysis
  • Topic modeling
  • Key insights extraction
  • Contextual relevance scoring

Quick Start

Prerequisites

• Python 3.11+
• Google API key (Get it here)

Document Placement

You can add documents in two ways:

1. Through the Web Interface

   • Use the drag & drop interface
   • Maximum 16MB per file

2. Manual Placement

   # Place PDF files directly in the docs folder
   project_root/
   └── docs/
       ├── document1.pdf
       ├── document2.pdf
       └── ...

   The system will automatically process any PDF files in the docs folder on startup.

Installation

1. Setup Environment

# Clone and enter directory
git clone <repository-url> && cd <project-folder>

# Create virtual environment
python -m venv venv
source venv/bin/activate  # Windows: venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

2. Configure API Key

# Create .env file in backend directory
echo "GOOGLE_API_KEY=your_key_here" > backend/.env

3. Run Application

# Start server
chmod +x start.sh && ./start.sh

# Open in browser
http://localhost:5001

Usage

1. Upload Documents

   • Drag & drop PDFs or click to browse
   • Max file size: 16MB per PDF
   • Supports text-based PDFs

2. Ask Questions

   • Type your question
   • Get answers with source citations
   • View AI-powered insights

Configuration

Key parameters can be adjusted in:

# document_processor.py
chunk_size = 800        # Text chunk size
chunk_overlap = 50      # Overlap between chunks

# rag_pipeline.py
k = 8                  # Number of chunks to retrieve

Troubleshooting

Common issues and solutions:

• Port in Use: Change port in start.sh or kill existing process
• PDF Errors: Ensure PDFs are text-based (not scanned)
• API Issues: Verify API key and quota limits

API Endpoints

POST /api/upload    # Upload PDF files
POST /api/ask      # Query documents
GET  /api/stats    # Get document statistics