# 💬 Comment Toxicity Classification App [](https://huggingface.co/spaces/ad-2004/comment-toxicity-analyser) [](https://www.python.org/) [](https://streamlit.io/) This deep learning-based application classifies text comments into various toxicity categories—including hate speech, offensive language, and harassment. Built to help moderate online platforms, it filters harmful content in real-time to ensure safer interactions. --- ## 📸 Demo Here's a quick look at the application in action:  --- ## ✨ Features - **Multi-Label Classification:** Detects `toxic`, `severe toxic`, `obscene`, `threat`, `insult`, and `identity hate` labels. - **Deep Learning Powered:** Uses an LSTM recurrent neural network for accurate contextual predictions. - **Interactive Interface:** Developed with Streamlit for a seamless and responsive user experience. - **Real-Time Processing:** Quickly analyzes and classifies user inputs on the fly. --- ## 🛠️ Tech Stack - **Frontend:** Streamlit - **Backend & AI:** Python, TensorFlow/Keras - **Data Processing:** Pre-trained text embeddings - **Storage:** Vocabulary files hosted on Supabase; model available via Google Drive --- ## 🚀 Getting Started Follow these steps to get the project running on your local machine. 1. **Clone the Repository:** ```bash git clone [https://github.com/your-repo/comment-toxicity-analyser.git](https://github.com/your-repo/comment-toxicity-analyser.git) cd comment-toxicity-analyser ``` 2. **Create and Activate a Virtual Environment:** ```bash # For macOS/Linux python3 -m venv venv source venv/bin/activate # For Windows python -m venv venv .\venv\Scripts\activate ``` 3. **Install Dependencies:** ```bash pip install -r requirements.txt ``` 4. **Download the Pre-trained Model:** The model is hosted on Google Drive. Please [download the `.h5` model file from this link](https://drive.google.com/file/d/19sR7F2iiyND4bhKaQDYNm5f-cN4c3nq_/view?usp=drive_link) and place it in the root directory of the project. 5. **Run the Application:** ```bash streamlit run app.py ``` --- ## 📝 How to Use After installation, run the application using `streamlit run app.py` and navigate to `http://localhost:8501`. Enter any comment or text into the input box and click "Analyze" to see the toxicity classification. --- ## 🎯 Project Context ### Challenges Solved This project accurately analyzes the context to detect nuanced toxicity, efficiently processes unstructured data including informal language and slang, and maintains balanced performance by reducing false positives and negatives. ### Future Improvements Plans for future enhancements include increasing model accuracy through domain-specific fine-tuning, adding multi-language support, and integrating sentiment analysis for deeper insights. ### Impact By automating toxicity detection, this tool contributes to healthier online interactions and significantly reduces the manual moderation workload on online platforms. --- ## 🤝 Contributing Contributions are welcome! If you have suggestions for improvements or find a bug, please feel free to open an issue or submit a pull request. # Dl_comment_toxicity