ml-poc-classification

A light-weight proof of concept (PoC) for text classification using a fine-tuned BERT model. The project demonstrates hate speech detection by classifying raw input text into pre-defined labels via a simple web interface.

---

📋 Project Overview

Minimal Machine Learning pipeline for BERT (bert-base-uncased)-based text classification. Includes training, evaluating, and a very small Flask-based web application for real-time inference.

Core Features

• Load CSV data and train a BERT-based classifier
• Evaluate classification performance
• Run a Flask server for real-time text classification
• Predicts the type of hate speech based on user input

This PoC is developed as a technical foundation for a future larger project that will include OCR integration, database support, and real-time communication via WebSockets.

---

📦 Installation & Setup

1️⃣ Clone the Repository

git clone https://github.com/NathanGrs00/ml-poc-classification.git
cd ml-poc-classification

2️⃣ Install Dependencies

Use pip to install all the required packages:

pip install -r requirements.txt

💡 Python 3.8+ is recommended.

3️⃣ Launch the Web Server

Start the Flask app:

python main.py

Click the localhost link flask generates. It will show a simple HTML page with an input field where you can type or paste text to classify.

---

🚀 Usage

1. Go to http://localhost:5000
2. Enter any text (e.g., a tweets or comments)
3. Click the "Submit" button
4. The model returns a label with the type of hate speech (if any)

---

🗺️ Roadmap

Feedback from peer students included the following changes to be implemented in the Proof of Concept:

• Model is not reliable enough.
  • Extend dataset with more data.
  • Balance the percentage of hateful and neutral comments.
  • Verify the current labeled data in the dataset.
• UI looks too simple, result is hard to read.
  • Change the labels to a more formatted version. Instead of just displaying 'violence' display a little more information.
  • Give each result a colored label

This PoC is a stepping stone towards a larger application. Future plans are:

• 🧾 Adding OCR to scan document/text image input
• 🗄️ Database connection for saving classification output and user feedback.
• 🔌 WebSockets for real-time updates

---

📣 Feedback & Contribution

I love feedback and contributions!

Report bugs or request features via GitHub Issues.

Fork the repository and make a Pull Request to contribute.

Feel free to propose improvements!

---

✅ License

This project is licensed under the MIT License.
See the LICENSE file for more information.

---

🤝 Contributing

Contributions are always welcome!

To contribute:

1. Fork the repository
2. Create a new branch for your feature or bugfix
3. Commit your changes with clear messages
4. Push to your fork
5. Open a Pull Request describing your changes

For major changes, consider opening an issue first to discuss your proposal.