EUVIMEDEUVIMED entdecken
Crossref

Trustworthy Phishing URL Detection Using Machine Learning: Performance Analysis and Future Directions

Omar Almomani, Ahmad Shalaldeh, Mohammed Amin Almaiah, Ahmad Saleh Ali Shatat, Mohammed Madi, Mutasem K. Alsmadi, Ibrahim Mohammed Dweib

International Journal of Advances in Soft Computing and its Applications · 2026

DOI: 10.15849/ijasca.129 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Journal of Advances in Soft Computing and its Applications, 2026. Verfasst von Omar Almomani und 6 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Over the years, Uniform Resource Locator (URL) Phishing has been a threat to online security because their scams exploit trust in a misleading URL to trick unsuspecting users into disclosing their username or passwords and other personal information. Not only that, but it is also crucial to identify these malicious URLs and ensure online platforms and users are protected. The ISCXURL2016 data was utilized for the purpose of analyzing the performance of different Machine Learning (ML) models in differentiating phishing and legitimate URLs in this paper. A wide range of classifiers were explored including the classic classifiers as well as more advanced ensemble classifiers. The results of the experiments indicate that ensemble-based boosting models can be more beneficial for URL phishing detection. HistGradientBoost achieved the highest accuracy of 98.47%, followed closely by LightGBM and XGBoost that both outperformed the accuracy of 98%. The other classifiers, RandomForest and ExtraTrees, also performed well with accuracy above 97% and traditional classifiers such as Quadratic Discriminant Analysis (QDA) showed the poorest performance with an accuracy of 67.75%. The results show that the boosting-based ensembles provide a suitable negation of the complex and non-linear structure of phishing URLs. Along with empirical evaluation, it offers a Properties & Positioning viewpoint as well as the importance of the ML-based phishing detection in the broader context of cybersecurity issues. Moreover, it discusses the future directions, with Federated Learning (FL) in the context of decentralized and privacy-preserving collaboration and Large Language Models (LLM) in the context of semantic reasoning and flexibility. These insights together will provide the building blocks for more intelligent, adaptive and trustworthy phishing organizations.

Zitieren

APA

Almomani, O., Shalaldeh, A., Almaiah, M. A., Shatat, A. S. A., Madi, M., Alsmadi, M. K., & Dweib, I. M. (2026). Trustworthy Phishing URL Detection Using Machine Learning: Performance Analysis and Future Directions. International Journal of Advances in Soft Computing and its Applications. https://doi.org/10.15849/ijasca.129

Vancouver

Almomani O, Shalaldeh A, Almaiah MA, Shatat ASA, Madi M, Alsmadi MK, et al. Trustworthy Phishing URL Detection Using Machine Learning: Performance Analysis and Future Directions. International Journal of Advances in Soft Computing and its Applications. 2026. doi:10.15849/ijasca.129

BibTeX
@article{Almomani2026,
  author = {Almomani, Omar and Shalaldeh, Ahmad and Almaiah, Mohammed Amin and Shatat, Ahmad Saleh Ali and Madi, Mohammed and Alsmadi, Mutasem K. and Dweib, Ibrahim Mohammed},
  title = {Trustworthy Phishing URL Detection Using Machine Learning: Performance Analysis and Future Directions},
  journal = {International Journal of Advances in Soft Computing and its Applications},
  year = {2026},
  doi = {10.15849/ijasca.129}
}
Kostenlos starten →