Study PDF Platform - RAG-ready

A comprehensive PDF study platform with AI-powered question answering using Cohere embeddings and Google Gemini for image analysis.

Features

• AI-Powered Q&A: Ask questions about PDF documents and get intelligent answers
• Multi-Image Analysis: Analyzes multiple relevant pages for comprehensive answers
• Thai Language Support: Optimized for Thai educational content with TA Tohtoh personality
• Modern UI: Responsive design with dark theme
• Real-time Chat: Interactive chat interface for document queries

Tech Stack

Backend

• Node.js with Express
• Cohere AI for embeddings
• Google Gemini for image analysis
• CORS enabled for frontend communication

Frontend

• Vanilla HTML, CSS, JavaScript
• Responsive design
• Real-time chat interface

Setup Instructions

1. Backend Setup

cd backend
npm install

2. Environment Variables

Create a .env file in the backend directory:

cp env.example .env

Edit .env and add your API keys:

CO_API_KEY=your_cohere_api_key_here
GEMINI_API_KEY=your_gemini_api_key_here

3. Start the Application

# Start the backend server
cd backend
npm start

# The application will be available at http://localhost:3000

API Endpoints

• POST /search-and-answer - Main endpoint for asking questions
• POST /search - Search for relevant images
• POST /answer - Answer questions about specific images
• GET / - Serves the frontend

Usage

1. Open your browser and go to http://localhost:3000
2. Ask questions about the PDF documents in Thai or English
3. The AI will analyze relevant pages and provide comprehensive answers

File Structure

├── backend/
│   ├── api.js                 # Main server file
│   ├── package.json           # Backend dependencies
│   ├── pdf_images/            # Processed PDF images
│   ├── pdf_image_embeddings.json  # Pre-computed embeddings
│   └── processed_image_paths.txt  # Image paths mapping
├── frontend/
│   ├── index.html             # Main frontend file
│   └── styles.css             # Styling
└── README.md                  # This file

Development

The platform is designed to work with pre-processed PDF images and embeddings. The backend loads these on startup and uses them for semantic search and question answering.

License

This project is part of a CEDT Final Project.