abrhaleyarefaine1997/meta-tigrinya-harmful-content-detector

Machine learning-based system for detecting harmful content (hate speech, offensive language, misinformation) in Tigrinya. Includes API, dashboard, and Streamlit app for scalable, data-driven content moderation aligned with real-world trust & safety challenges. # Meta Tigrinya Harmful Content Detector ## Overview This project is a machine learning system designed to detect harmful content (hate speech, offensive language) with a focus on Tigrinya language. ## Components - **app/**: Streamlit application for user interaction - **dashboard/**: Visualization dashboard - **api/**: Backend API for inference - **data/**: Training and testing datasets - **models/**: Saved ML models - **notebooks/**: Experiments and research ## Features - Text classification for harmful content - Tigrinya-focused dataset - API + UI integration - Extendable for multilingual moderation ## Tech Stack - Python - NLP / Machine Learning - Streamlit - SQL ## Future Improvements - Transformer models (BERT, multilingual) - Better Tigrinya dataset - Real-time moderation pipeline ## Author Abrhaley