The PhishBench platform extracts features from emails and URLs, run classifiers, and evaluates them using evaluation metrics.

Installation

PhishBench works on Python 3.7 and 3.8.

To install the latest version on master, run

pip install git+https://github.com/ReDASers/Phishing-Detection.git

To install a specific version, run

pip install git+https://github.com/ReDASers/Phishing-Detection.git@{ version number }

For an example, to install version 1.1.4, run

pip install git+https://github.com/ReDASers/[email protected]

How to run a PhishBench Basic Experiment

The PhishBench Basic experiment script performs an experiment with the following workflow:

1. Load the dataset
2. Extract features
3. Pre-process features
4. Train classifiers on a training set
5. Evaluate the trained classifiers on a held-out test set

Make a config file

The basic experiment is controlled by a configuration file. To create a starter config file, run the make-phishbench-config command.

usage: make-phishbench-config [-h] [-v] [-f CONFIG_FILE]

PhishBench Config Generator

optional arguments:
  -h, --help            show this help message and exit
  -v, --verbose         Increase output verbosity
  -f CONFIG_FILE, --config_file CONFIG_FILE
                        The name of the config file to generate.

Example:

make-phishbench-config -f Test.ini

Edit the configuration file in a text editor and save it. Use the documentation for guidance.

Run Experiment

Once you have your configuration file, you can us the phishbench command to run the basic experiment.

usage: phishbench [-h] [--version] [-f CONFIG_FILE] [-v] [-o OUTPUT_INPUT_DIR]
                  [-c]

PhishBench Basic Experiment Script

optional arguments:
  -h, --help            show this help message and exit
  --version             Display the PhishBench version number and exit
  -f CONFIG_FILE, --config_file CONFIG_FILE
                        The config file to use
  -v, --verbose         Increase output verbosity
  -o OUTPUT_INPUT_DIR, --output_input_dir OUTPUT_INPUT_DIR
                        Output/input directory
  -c, --ignore_confirmation
                        Do not wait for user's confirmation

Use PhishBench as a library

For experiments which deviate from the basic experiment workflow, you can also use PhishBench as a library. To do so, import phishbench and the desired modules in your script and call the initialize function.

import phishbench

phishbench.initalize('Config_File.ini')

Citation

@inproceedings{
    10.1145/3372297.3420017,
    author = {Zeng, Victor and Zhou, Xin and Baki, Shahryar and Verma, Rakesh M.},
    title = {PhishBench 2.0: A Versatile and Extendable Benchmarking Framework for Phishing},
    year = {2020},
    isbn = {9781450370899},
    publisher = {Association for Computing Machinery},
    address = {New York, NY, USA},
    url = {https://doi.org/10.1145/3372297.3420017},
    doi = {10.1145/3372297.3420017},
    abstract = {We describe version 2.0 of our benchmarking framework, PhishBench. With the addition of the ability to dynamically load features, metrics, and classifiers, our new and improved framework allows researchers to rapidly evaluate new features and methods for machine-learning based phishing detection. Researchers can compare under identical circumstances their contributions with numerous built-in features, ranking methods, and classifiers used in the literature with the right evaluation metrics. We will demonstrate PhishBench 2.0 and compare it against at least two other automated ML systems.},
    booktitle = {Proceedings of the 2020 ACM SIGSAC Conference on Computer and Communications Security},
    pages = {2077–2079},
    numpages = {3},
    keywords = {machine learning, phishing, automatic framework, benchmarking},
    location = {Virtual Event, USA},
    series = {CCS '20}
}