Published Online:January 2026
Product Name:The IUP Journal of Computer Sciences
Product Type:Article
Product Code:IJCS020126
DOI:10.71329/IUPJCS/2026.20.1.17-26
Author Name:Aderibigbe Malik Mayowa and Francisca Nonyelum Ogwueleka
Availability:YES
Subject/Domain:Engineering
Download Format:PDF
Pages:17-26
The paper provides a comprehensive overview of machine learning (ML) techniques for malicious uniform resource locators (URL) detection, emphasizing both traditional and modern approaches. Malicious URLs are a major threat to online security, often leading to malware infections, phishing attacks, and financial fraud. Despite the widespread use of blacklisting methods, these are increasingly ineffective as attackers evolve new tactics to bypass detection systems. The paper focuses on the potential of ensemble ML techniques, such as AdaBoost, Bagging, Stacking, and Voting, to improve detection accuracy and overcome the limitations of traditional models. The ensemble methods, which combine the strengths of various classifiers, have shown promise in addressing issues such as overfitting, bias-variance tradeoff, and performance inconsistencies. The paper provides a detailed comparative analysis of ensemble algorithms and explores their role in enhancing the robustness and scalability of malicious URL detection systems. It concludes with a discussion on future research directions for improving URL detection using ML.
Malicious Uniform Resource Locators (URLs) represent a significant threat to online security, primarily due to their involvement in phishing, malware distribution, and other cyberattacks. With the proliferation of web applications across various sectors, malicious actors have increasingly targeted vulnerable websites to exploit users for financial gain, identity theft, and other illicit activities (Hu et al., 2020).