2026-04-17T17:45:25.680Z http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai

ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai:3:9956 2025-06-13T13:58:39Z 3

Using BERT with Modified Metaheuristic Optimized XGBoost for Phishing Email Identification, Chapter in ISEM Information Systems Engineering and Management: ICAIS 2024: International Conference on Artificial Intelligence and Smart Energy, Springer, volume 4 2024 http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai/record/3/9956 https://link.springer.com/chapter/10.1007/978-3-031-61475-0_28 M. Antonijevic L. Jovanovic N. Bacanin M. Zivkovic J. Kaljevic T. Zivkovic This research investigates the potential of Natural Language Processing (NLP) in discerning the underlying motivations behind phishing emails. Traditional systems, which primarily focus on tracking origin points, subjects, and IP addresses, face challenges in detecting emerging networking threats that involve the virtual relocation of large-scale scam operations. Despite these challenges, the consistency of intent within the email context remains a constant factor. Therefore, this study proposes the utilization of NLP in conjunction with robust classification techniques as a promising approach to effectively address phishing emails and fortify cybersecurity. The specific focus of the study lies in exploring the application of BERT encoding combined with the XGBoost classifier for the identification of phishing emails based on their email body content. However, the success of machine learning classifiers is contingent upon appropriate hyperparameter selection. To tackle this issue, an altered version of the SCHO algorithm is introduced and customized to overcome inherent limitations, providing a valuable benchmark against other contemporary optimizers. Simulations conducted with real-world samples demonstrate promising outcomes, achieving a precision rate exceeding 75% for phishing email identification. bookPart Springer, Cham 358 370 10.1007/978-3-031-61475-0_28 ISEM Information Systems Engineering and Management: ICAIS 2024: Proceedings of 4th International Conference on Artificial Intelligence and Smart Energy, volume 4