BytePhilosopher/Amharic-sentiment-analysis

Amharic sentiment analysis is a machine learning tool that classifies Amharic text into positive, negative, or neutral sentiment. It helps analyze public mood in low-resource languages using simple, effective models. ```markdown # 🇪🇹 QalAnalyzer (ቃል Analyzer) **QalAnalyzer** is a machine learning-based sentiment analysis tool for the Amharic language. It classifies Amharic text into three categories: **Positive**, **Negative**, and **Neutral**. Built with a focus on low-resource language NLP, this project uses classical ML methods and a simple web app interface for practical use in Ethiopia and beyond. --- ## 🌟 Features - 🔤 Amharic text preprocessing and cleaning - 📊 TF-IDF feature extraction - 🤖 Sentiment classification using Logistic Regression - 🧪 Model evaluation and performance reports - 🌐 Streamlit web app for real-time sentiment analysis - 📁 Modular code and organized project structure --- ## 📂 Project Structure ``` qal-analyzer/ │ ├── data/ # Raw and cleaned datasets ├── notebooks/ # Jupyter notebooks (EDA, training, evaluation) ├── models/ # Saved models (TF-IDF, Logistic Regression) ├── app/ # Streamlit app ├── utils/ # Text cleaning utilities ├── README.md ├── requirements.txt └── LICENSE ```` --- ## 📊 Dataset We used the publicly available LiyaSileshi Amharic Sentiment Dataset, containing over 9,000 Amharic tweets labeled as: - Positive - Negative - Neutral --- ## 🛠️ Tech Stack | Purpose | Tool/Library | |----------------------|-------------------------| | Programming Language | Python | | ML Framework | Scikit-learn | | Web App | Streamlit | | Visualization | Matplotlib, Seaborn | | Text Processing | NLTK, Regex | | Deployment Ready | GitHub + Streamlit | --- ## 🚀 How to Run Locally 1. **Clone the repo:** ```bash git clone github.com cd qal-analyzer ```` 2. **Install dependencies:** ```bash pip install -r requirements.txt ``` 3. **Run the app:** ```bash streamlit run app/streamlit_app.py ``` --- ## 🧠 Model Overview * ** …