Rag-Model

RAG Document Analysis System

A comprehensive RAG (Retrieval-Augmented Generation) based document analysis system for analyzing disaster management training data. Built for the National Disaster Management Authority (NDMA) as part of Smart India Hackathon 2025.

🌟 Features

• 📤 Multi-Format Upload: Support for PDF, Excel (.xlsx, .xls), and CSV files
• 🤖 AI-Powered Analysis: Uses Google Gemini AI for intelligent data analysis
• 📊 Interactive Dashboard: Beautiful visualizations with charts and metrics
• 📄 Government Reports: Auto-generate professional PDF reports in government format
• 🔍 RAG-based Q&A: Ask questions about your uploaded documents
• 💾 Data Persistence: MongoDB storage for analysis history
• 🎨 Modern UI: Glassmorphism design with smooth animations

🏗️ Architecture

┌─────────────────┐
│  React Frontend │
│   (Vite + UI)   │
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│  Express API    │
│  (Node.js)      │
└────────┬────────┘
         │
    ┌────┴────┐
    ▼         ▼
┌────────┐ ┌──────────┐
│MongoDB │ │Gemini AI │
└────────┘ └──────────┘

🚀 Quick Start

Prerequisites

• Node.js (v18 or higher)
• MongoDB (local or Atlas)
• Google Gemini API Key

Installation

1. Clone the repository

cd "c:\Users\91979\OneDrive\Desktop\Rag Model"

1. Setup Backend

cd backend
npm install

1. Configure Environment Variables

Edit backend/.env and add your credentials:

GEMINI_API_KEY=your_gemini_api_key_here
MONGODB_URI=mongodb://localhost:27017/rag-document-analysis

1. Setup Frontend

cd ../frontend
npm install

Running the Application

1. Start MongoDB (if running locally)

mongod

1. Start Backend Server

cd backend
npm run dev

Backend will run on http://localhost:5000

1. Start Frontend (in a new terminal)

cd frontend
npm run dev

Frontend will run on http://localhost:5173

1. Open in Browser Navigate to http://localhost:5173

📖 Usage Guide

1. Upload Training Data

• Click on the upload area or drag & drop your file
• Supported formats: PDF, Excel, CSV
• Maximum file size: 10MB
• Click "Upload & Analyze"

2. View Analysis Dashboard

• Automatic analysis using AI
• View key metrics (trainings, participants, completion rates)
• Interactive charts for theme and state distribution
• Gap analysis showing underserved areas
• Key insights and recommendations

3. Download Report

• Click on "Report" tab
• Preview the government-format report
• Download as PDF
• Share with stakeholders

📁 Project Structure

Rag Model/
├── backend/
│   ├── models/
│   │   └── Document.js          # MongoDB schema
│   ├── routes/
│   │   ├── upload.js            # Upload endpoints
│   │   └── analysis.js          # Analysis endpoints
│   ├── services/
│   │   ├── documentParser.js    # PDF/Excel/CSV parser
│   │   ├── ragEngine.js         # RAG implementation
│   │   ├── analysisService.js   # AI analysis logic
│   │   └── reportGenerator.js   # PDF report generation
│   ├── server.js                # Express server
│   └── package.json
│
├── frontend/
│   ├── src/
│   │   ├── components/
│   │   │   ├── FileUpload.jsx   # Upload component
│   │   │   ├── AnalysisDashboard.jsx  # Dashboard
│   │   │   └── ReportPreview.jsx      # Report preview
│   │   ├── App.jsx              # Main app
│   │   ├── main.jsx             # Entry point
│   │   └── index.css            # Styles
│   ├── index.html
│   ├── vite.config.js
│   └── package.json
│
└── README.md

🔧 API Endpoints

Upload & Analysis

• POST /api/upload - Upload and analyze document
• GET /api/upload/documents - Get all documents
• GET /api/upload/document/:id - Get specific document
• DELETE /api/upload/document/:id - Delete document

Analysis & Reports

• POST /api/analysis/ask - Ask question about document (RAG)
• GET /api/analysis/report/:filename - Download report
• GET /api/analysis/stats - Get aggregate statistics

🎨 Tech Stack

Backend:

• Node.js + Express
• MongoDB + Mongoose
• Google Gemini AI
• PDFKit (report generation)
• Multer (file uploads)
• pdf-parse, xlsx, csv-parser

Frontend:

• React 18
• Vite
• Recharts (data visualization)
• React Dropzone
• Lucide React (icons)
• Axios

📊 Sample Data Format

Excel/CSV Format

Training ID,Date,Location,State,Theme,Participants,Trainer,Duration,Completion Rate
TR001,2024-01-15,Delhi,Delhi,Earthquake,50,Dr. Sharma,2 days,95%
TR002,2024-01-20,Mumbai,Maharashtra,Flood,75,Mr. Patel,3 days,88%

Analysis Output

{
  "totalTrainings": 150,
  "totalParticipants": 5000,
  "themeDistribution": {
    "Earthquake": 40,
    "Flood": 60,
    "Cyclone": 30
  },
  "stateWiseCoverage": {
    "Delhi": 20,
    "Maharashtra": 35
  },
  "averageCompletionRate": "91%",
  "gapAnalysis": {
    "underservedStates": ["Nagaland", "Mizoram"],
    "underservedThemes": ["Tsunami"]
  },
  "recommendations": [...]
}

🔐 Environment Variables

🤝 Contributing

This project was developed for Smart India Hackathon 2025 by Team JARVIS GGV.

📝 License

MIT License - Feel free to use for educational and government purposes.

👥 Team JARVIS GGV

Smart India Hackathon 2025
Problem Statement ID: SIH25258
Theme: Disaster Management

---

For Official Use by NDMA, Government of India