AI: Deep Learning for Phishing URL Detection

Model Performance

ROC/AUC Curve
Confusion Matrix
F1 Score

Requirements

This code was created with Python 3.6.7. Other versions of Python 3 might also work. You can have multiple Python
versions (2.x and 3.x) installed on the same system without problems.

Make sure to install all requirements:

$ pip install -r requirements.txt

NOTICE : Because of Github size limits, please download the model from here: https://www.zpettry.com/bi-lstmchar256256128.h5

Quick start

Ensure the model has been downloaded from the above link.

Open a separate tab or window and run:

$ python3 flaskrestapi.py

Now go back to the original tab or window and run:

$ python3 request.py -u https://www.google.com/about
Output:
$ [{'malicious percentage': 2.552182786166668, 'result': 'URL is probably NOT malicious.', 'url': 'https://www.google.com/about'}]

Web site and documentation

Blog and additional information about this project is available at the web site:

https://www.zpettry.com/

License

This code is licensed under the terms of the MIT License (see the file
LICENSE).