📧 Email Phishing Detection

Detect phishing emails in real-time using machine learning — trained on six merged datasets and deployed via a full-stack FastAPI app.

🔗 Live Demo: https://phishingdetection.net

📚 Table of Contents

• What It Does
• Tech Stack
• ML Models
• Sample Input
• Future Improvements
• Project Structure
• Acknowledgements
• Author

---

🧠 What It Does

This project allows users to paste real email content (sender, subject, body, etc.) and choose between three machine learning models to detect whether it's Phishing or Legitimate.

It combines:

• 📊 Natural Language Processing (TF-IDF + NLTK)
• 🤖 ML models (Naive Bayes, Logistic Regression, SVM)
• 🌍 Live API deployment + responsive UI

---

🛠️ Tech Stack

Layer	Tech
Frontend	HTML, CSS, JavaScript
Backend	FastAPI, Uvicorn
ML/NLP	Scikit-learn, NLTK, joblib
Deployment	Render, Namecheap
Hosting Models	Hugging Face 🤗

---

🤖 ML Models

Choose between:

• ✅ Support Vector Machine (Best Accuracy)
• ✅ Logistic Regression
• ✅ Multinomial Naive Bayes

📦 Hugging Face Models

• 🔗 SVM Model
• 🔗 TF-IDF Vectorizer

---

🧪 Sample Input

Sender: freeiphone@gmail.com
Subject: Don't miss this chance!
Body: Click the link to claim your free iPhone 14 Pro Max.
Date: May 5, 2025
Model: SVM

✔️ Output: Phishing

---

🚀 Future Improvements

• 📡 Public API & Documentation
  Provide a proper REST API endpoint with OpenAPI/Swagger documentation so developers can integrate the phishing detection system into their own applications.

• 🎨 Improve the UI
  Rebuild the frontend using a modern framework like React (possibly with Tailwind or Material UI) to create a more interactive and responsive user experience.

• 🔗 URL-Based Model
  Train and integrate a secondary model focused specifically on analyzing URLs for phishing characteristics such as domain structure, length, obfuscation, and suspicious keywords.

• 📈 Expand the Dataset
  Enhance the model's performance by collecting a larger and more diverse dataset of phishing and legitimate emails, improving generalization and reducing bias.

• 🧠 Improve Model Explainability
  Integrate explainable AI tools like SHAP or LIME to provide transparency into why the model classified an email as phishing or legitimate.

• 📬 Real-Time Email API Integration (Optional)
  Integrate with email providers like Gmail or Microsoft Outlook via API to allow live scanning of user inboxes (with permission) and flag suspicious messages in real time.

---

🗂️ Project Structure

EmailPhishingDetection/
├── api/
│   ├── main.py
│   └── pipeline.py
├── data/
│   └── phishing_email.csv
├── frontend/
│   └── static/
│       └── index.html
├── models/
│   ├── logistic_regression_model.joblib
│   ├── naive_bayes_model.joblib
│   ├── svm_model.joblib
│   └── tfidf_vectorizer.joblib
├── notebooks/
│   └── 01_training.ipynb
├── images/
├── .gitignore
├── LICENSE.md
├── README.md
├── render.yml
└── requirements.txt

---

🙏 Acknowledgements

This project uses the Phishing Email Dataset by Naser Abdullah Alam on Kaggle.

Please cite the following article if using this dataset:

Al-Subaiey, A., Al-Thani, M., Alam, N. A., Antora, K. F., Khandakar, A., & Zaman, S. A. U. (2024, May 19).
Novel Interpretable and Robust Web-based AI Platform for Phishing Email Detection.
ArXiv: https://arxiv.org/abs/2405.11619

---

👨‍💻 Author

Emre OTU
🔗 GitHub | LinkedIn