Smart Detection of Malicious URLs Using a Hybrid Machine Learning Approach

As digital technologies are advancing at a very fast rate, network security threats have also become more common. Cyber attackers use several techniques, including phishing, social engineering, and pharming, to take advantage of weaknesses and mislead users. A key part of these attacks is the employment of misleading Uniform Resource Locators (URLs) to trick people into visiting malicious websites. To mitigate this emerging threat, scholars have thoroughly examined machine learning and deep learning-based techniques for malicious URL detection. In this work, we present a sophisticated malicious URL detection methodology that utilizes machine learning methods based on distinctive URL features and behavioral patterns. We also incorporate big data analytics to improve detection performance by detecting suspicious URL anomalies in real time. Our detection system includes state-of-the-art URL-based feature extraction, a stable classification model, and highly scalable big data processing. Experimental results verify that our design improves the accuracy and efficiency of malicious URL detection significantly. The observations reflect the capability of our system as an efficient and scalable approach to reducing cyber threats from malicious URLs