Naomi-NLP/Yoruba-_Proverb-_Classification

# 🌟 OWE-YOR: Leveraging Transformer-Based Models for Yoruba Proverb Classification **OWE-YOR** is a state-of-the-art NLP project that focuses on the classification of **Yoruba proverbs** using Transformer-based models. It distinguishes between proverbs and non-proverbs, leveraging modern AI to preserve Yoruba language and culture. --- ## 🚀 Project Workflow - Introduction - Features - Dataset - Models - Usage - Demo - Results - Poster - Future Work - Team --- ## 📝 Introduction Yoruba proverbs are rich in cultural and linguistic meaning. Automating their detection is challenging due to limited annotated data. **OWE-YOR** uses Transformer-based models like **AfroLM** and **mBERT** to capture semantic and contextual subtleties, outperforming traditional machine learning models like Naive Bayes. **Goals of this project:** - Build an automated Yoruba proverb detection system ✅ - Compare traditional ML and Transformer-based approaches ⚡ - Contribute to AI tools for African languages 🌍 --- ## ✨ Features - Detects Yoruba proverbs vs. non-proverbs - Uses **Transformer-based models** for high accuracy - Provides a **Streamlit demo** for interactive testing - Easily extensible to other low-resource African languages --- ## 📂 Dataset - **Proverbs:** labeled `1` - **Non-proverbs:** labeled `0` - Data is **cleaned and preprocessed** for optimal training. > Dataset link: [huggingface.co] --- ## 🤖 Models - **Multinomial Naive Bayes** – baseline performance - **AfroLM** and **mBERT** – Transformer-based classification **Performance Comparison Example:** | Model | Accuracy | |-------------------------|---------| | Multinomial Naive Bayes | 85% | | AfroLM / mBERT | 92–95% | > Transformers better capture semantic context in Yoruba proverbs 🌟 --- ## 🌐 Demo Try the web app live on Hugging Face Spaces: 🎯 OWE-YOR: Yoruba Proverb Classifier --- ## 📊 Results - High accuracy in classifying Yoruba prove …