AI Text Detection Tool

A modular, full-stack system for detecting AI-generated text in news and academic essays.
Supports API, dashboards, web apps, and a Chrome extension for real-time scanning and explainability.

---

Table of Contents

• AI Text Detection Tool
  • Table of Contents
  • Features
  • Prerequisites
  • Installation
  • Configuration
  • Running the Project
    • Backend API
    • Dashboard (Plotly Dash)
    • React Frontend
    • Chrome Extension
  • API Usage
  • Project Structure
  • Contributing
  • Roadmap
  • Troubleshooting
  • License
  • Contact
  • Acknowledgments

---

Features

• FastAPI Backend with a fine-tuned RoBERTa model for high-accuracy inference.
• LIME Explainability highlights which words influenced each prediction.
• Plotly Dash Dashboard for interactive EDA, model evaluation, and inference.
• React Web App for single-text and batch-file analysis with a responsive UI.
• Chrome Extension enabling on-page scanning of any web content.

---

Prerequisites

• Python 3.10 or higher
• Node.js 14.x or higher + npm
• Optional: Docker & Docker Compose (for containerized setup)

---

Installation

1. Clone the repo

   git clone https://github.com/MichaelShpyl/AI-Text-Detection-Tool.git
   cd AI-Text-Detection-Tool

2. Create & activate a Python virtual environment

   python -m venv venv
   source venv/bin/activate      # Linux/macOS
   venv\Scripts\activate.bat   # Windows

3. Install Python dependencies

   pip install --upgrade pip
   pip install -r requirements.txt

4. Install frontend dependencies

   cd frontend
   npm install
   cd ..

---

Configuration

• .env file (optional): You can create a .env in the project root with:

  BACKEND_HOST=127.0.0.1
  BACKEND_PORT=8000
  MODEL_PATH=./models/roberta_ai_detector.pt

• Dash settings: By default, the Dash app will connect to http://127.0.0.1:8000. Adjust in scripts/dashboard_app.py if needed.

---

Running the Project

Backend API

cd scripts
python api_server.py

• Starts FastAPI at http://127.0.0.1:8000
• Loads the trained model into memory

---

Dashboard (Plotly Dash)

• Option A – Notebook
  Open notebooks/06_dashboard_dash.ipynb in Jupyter and Run All.
• Option B – Script

  python scripts/dashboard_app.py

  • Dashboard runs at http://localhost:8050
  • Tabs: EDA, Evaluation, Inference (API-backed)

---

React Frontend

cd frontend
npm start

• Opens at http://localhost:3000
• Sidebar: Single Text Analysis | Batch Upload

---

Chrome Extension

1. Go to chrome://extensions/, enable Developer mode
2. Click Load unpacked, select the extension/ folder
3. Visit any webpage, click Scan this article
4. View predictions & highlights (ensure API is running)

---

API Usage

# Single-text inference
curl -X POST http://127.0.0.1:8000/predict   -H "Content-Type: application/json"   -d '{"text": "Sample input text to analyze."}'

{
  "prediction": "AI-generated",
  "confidence": 0.97
}

---

Project Structure

AI-Text-Detection-Tool/
├── scripts/                 # Backend API & Dash launcher
│   ├── api_server.py
│   └── dashboard_app.py
├── notebooks/               # Jupyter analysis & Dash notebook
├── frontend/                # React single/batch analysis app
├── extension/               # Chrome extension source code
├── data/                    # Sample datasets for batch testing
├── models/                  # Trained model weights
├── requirements.txt         # Python dependencies
├── package.json             # Frontend dependencies
├── README.md                # This file
└── LICENSE                  # MIT license

---

Contributing

1. Fork the repo
2. Create a feature branch: git checkout -b feature/awesome-new
3. Commit your changes: git commit -m "feat: add awesome feature"
4. Push to your branch: git push origin feature/awesome-new
5. Open a Pull Request

Please follow the Contributing Guidelines if available.

---

Roadmap

• Add automated unit tests & CI via GitHub Actions
• Containerize with Docker & Docker Compose
• Publish Python package to PyPI
• Enhance Chrome extension UI with dark mode toggle

---

Troubleshooting

• API errors? Check scripts/api_server.py logs for stack traces.
• Model loading issues? Verify MODEL_PATH points to the correct .pt file.
• Port conflicts? Change ports via .env or command-line flags.

---

License

This project is licensed under the MIT License – see the LICENSE file for details.

---

Contact

Michael Shpyl – michael.shpyl@gmail.com
Project repo: https://github.com/MichaelShpyl/AI-Text-Detection-Tool

---

Acknowledgments

- Built with ❤ using FastAPI, Plotly Dash, React, and LIME

• Thanks to DSP Sligo Business Improvement Team for guidance