Ayamgenerationalthinker/Ghana_fake_news_detector

A machine learning project to detect fake news in Ghana using NLP # 🔍 Ghana Fake News Detector A machine learning-powered Streamlit web app that detects fake news articles, built specifically for Ghana's media space. ## 🚀 Features - Classifies news as `FAKE`, `REAL`, or `UNCERTAIN` - Interactive Streamlit interface with modern UI - Confidence score visualization - Batch analysis for multiple headlines - URL-based article analysis - Extracts key features like word count, capital letters, numbers, etc. - Built with `scikit-learn`, `Streamlit`, `Plotly`, and `pandas` ## 📦 Installation ```bash git clone github.com cd Ghana_fake_news_detector python -m venv venv venv\Scripts\activate # For Windows pip install -r requirements.txt ▶️ Run the App bash Copy Edit streamlit run app.py The app will open in your browser at localhost. 🧠 Model Details Model: Logistic Regression Text vectorization: TF-IDF (max 5000 features) Trained on labeled Ghanaian news (fake & real) Supports custom training and fine-tuning 📚 Sources & Datasets GhanaWeb, JoyNews, CitiNews, Dubawa Ghana fact-checks Fake news headlines from social media and verified misinformation posts 🔧 File Structure graphql Copy Edit Ghana_fake_news_detector/ │ ├── app.py # Main Streamlit app ├── models/ # Saved TF-IDF and classifier model ├── data/ # Optional: training dataset ├── utils.py # Helper functions ├── requirements.txt # All dependencies └── README.md # This file 👨‍💻 Author Prince Ofosu Fiebor (AyamGenerationalThinker) Digital Skills Advocate | Tech Educator | Data & AI Enthusiast 🌐 Live Deployment (optional) You can deploy this app on: Streamlit Cloud Render Hugging Face Spaces 🙌 Contributions Pull requests are welcome! For major changes, please open an issue first.