Using Chat GPT for Malicious Web Links Detection

Thomas Kaisser, Claudia-Ioana Coste

2024

Abstract

Over the last years, the Internet has monopolized most businesses and industries. These outstanding advancements lead to the dangerous development of specialized threats employed to outsmart everyday users, collect personal data and financial benefits. One of the most relevant attacks is malicious web links, which can be inserted into private messages, emails, social media posts and others to deceive consumers and trick them into clicking. Present approach will classify links based on multiple manually extracted features. Then, we perform a feature importance analysis. Moreover on a smaller dataset, we employ OpenAI’s models to classify and then add a new feature representing the Chat GPT classification. Thus, we manage to improve the overall performance of multiple machine learning methods. The first experiment considers only a Random Forest classifier but in the second one, we added thirteen other intelligent algorithms and ensembles constructed from the best performing ones. The best obtained accuracy (95%) is reached by the RF model on the whole dataset.

Paper Citation

in Harvard Style

Kaisser T. and Coste C. (2024). Using Chat GPT for Malicious Web Links Detection. In Proceedings of the 20th International Conference on Web Information Systems and Technologies - Volume 1: WEBIST; ISBN 978-989-758-718-4, SciTePress, pages 425-432. DOI: 10.5220/0013069200003825

in Bibtex Style

@conference{webist24,
author={Thomas Kaisser and Claudia-Ioana Coste},
title={Using Chat GPT for Malicious Web Links Detection},
booktitle={Proceedings of the 20th International Conference on Web Information Systems and Technologies - Volume 1: WEBIST},
year={2024},
pages={425-432},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013069200003825},
isbn={978-989-758-718-4},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 20th International Conference on Web Information Systems and Technologies - Volume 1: WEBIST
TI - Using Chat GPT for Malicious Web Links Detection
SN - 978-989-758-718-4
AU - Kaisser T.
AU - Coste C.
PY - 2024
SP - 425
EP - 432
DO - 10.5220/0013069200003825
PB - SciTePress