A Dynamic Anomaly Detection Framework for Malicious URL Identification

The rapid expansion of web-based services has brought along an increase in the risks associated with malicious URLs. Conventional detection mechanisms often rely on predefined features, which may not effectively capture the dynamic nature of modern cyber threats. This paper introduces a Dynamic Anomaly Detection Framework (DADF) for identifying malicious URLs by integrating clustering algorithms with behavioral pattern analysis. The proposed system dynamically detects deviations from normal URL behavior within clusters, allowing for the identification of new and evolving threats. Experimental results show that DADF surpasses traditional machine learning models like Support Vector Machines (SVM) and Logistic Regression in terms of detection accuracy and adaptability.