🚀 Youtube Toxicity Analysis App

This project is a YouTube comment analysis app that scrapes comments from a given YouTube video URL, classifies them according to a toxicity model, and provides an analysis of the comments based on their toxicity levels. The toxicity model utilized in this project is a Bidirectional LSTM model, and a vectorizer model is used to vectorize the text data.

See Demo 📺 Click here

Features:

• Scrapes comments from a YouTube video URL.
• Classifies comments based on toxicity using a Bidirectional LSTM model.
• Provides analysis and visualization of comment toxicity levels.

Startup Guide 🚀

1. Clone this Repository on your local machine
2. Create a virtual environment conda create -n venv python=3.8
3. Activate it conda activate venv
4. Install initial deps pip install Requirements.txt
5. Iniate Model training by python model_trainer.py
6. Get Youtube Data Api v3 API_KEY from Google cloud console for free
7. Run the app python app.py

Models:

• Toxicity Model: Bidirectional LSTM model trained to classify comment toxicity levels.
• Vectorizer Model: Used to vectorize the text data for input to the toxicity model.

Technologies Used:

• Python
• Streamlit
• TensorFlow
• googleapiclient.discovery (for youtube comment scraping)

Credits :

This project was developed by Mohamed Hmida.

License:

This project is licensed under the MIT License.