aryanraj6308/ai-phising-email-detector

# AI Phishing Email Detector A machine learning web app that predicts whether an email or message is likely to be phishing. The project uses Python, scikit-learn, and Streamlit, and is structured so it can be uploaded directly to GitHub. ## Features - Paste or upload email text for analysis - Predicts phishing probability with a trained NLP model - Highlights suspicious words and phrases - Shows confidence score, risk level, and safety tips - Includes a small sample dataset for quick local testing - Supports replacing the sample data with larger Kaggle phishing datasets ## Tech Stack - Python - Streamlit - scikit-learn - pandas - joblib ## Project Structure ai-phishing-email-detector/ ├── app.py ├── train_model.py ├── requirements.txt ├── README.md ├── LICENSE ├── .gitignore ├── data/ │ └── sample_emails.csv ├── models/ │ └── .gitkeep └── src/ ├── __init__.py ├── detector.py └── suspicious_terms.py ## Setup 1. Create and activate a virtual environment: python -m venv .venv .venv\Scripts\activate 2. Install dependencies: pip install -r requirements.txt 3. Train the model: python train_model.py 4. Run the dashboard: streamlit run app.py ## Using a Kaggle Dataset The bundled dataset is intentionally small so the project works immediately. For better accuracy, download a phishing email or SMS dataset from Kaggle and format it as a CSV with these columns: text,label "Your email message here",phishing "A normal safe message here",safe Then train with: python train_model.py --data path/to/your_dataset.csv Labels can be `phishing` / `safe`, `1` / `0`, or similar common values. ## How It Works ## Important Note This tool is for learning and portfolio purposes. It should not be used as the only security control for real email systems.

标签：后端开发