PhishGNN: A Phishing Website Detection Framework using Graph Neural Networks

Tristan Bilot, Grégoire Geis, Badis Hammi

2022

Abstract

Because of the importance of the web in our daily lives, phishing attacks have been causing a significant damage to both individuals and organizations. Indeed, phishing attacks are today among the most widespread and serious threats to the web and its users. Currently, the main approaches deployed against such attacks are blacklists. However, the latter represent numerous drawbacks. In this paper, we introduce PhishGNN, a Deep Learning framework based on Graph Neural Networks, which leverages and uses the hyperlink graph structure of websites along with different other hand-designed features. The performance results obtained, demonstrate that PhishGNN outperforms state of the art results with a 99.7% prediction accuracy.

Paper Citation

in Harvard Style

Bilot T., Geis G. and Hammi B. (2022). PhishGNN: A Phishing Website Detection Framework using Graph Neural Networks. In Proceedings of the 19th International Conference on Security and Cryptography - Volume 1: SECRYPT, ISBN 978-989-758-590-6, pages 428-435. DOI: 10.5220/0011328600003283

in Bibtex Style

@conference{secrypt22,
author={Tristan Bilot and Grégoire Geis and Badis Hammi},
title={PhishGNN: A Phishing Website Detection Framework using Graph Neural Networks},
booktitle={Proceedings of the 19th International Conference on Security and Cryptography - Volume 1: SECRYPT,},
year={2022},
pages={428-435},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0011328600003283},
isbn={978-989-758-590-6},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 19th International Conference on Security and Cryptography - Volume 1: SECRYPT,
TI - PhishGNN: A Phishing Website Detection Framework using Graph Neural Networks
SN - 978-989-758-590-6
AU - Bilot T.
AU - Geis G.
AU - Hammi B.
PY - 2022
SP - 428
EP - 435
DO - 10.5220/0011328600003283