Masked and Phishing URL Detection using Machine Learning

Abstract

With the escalating threats in the digital landscape of cybersecurity, the rapid and widespread of masked and phishing URLs poses a significant threat to online users. Detecting these malicious URLs is a paramount concern to safeguard sensitive information and prevent unauthorized access. This study delves into the application of machine learning algorithms for the accurate identification of masked and phishing URLs. Specifically, Decision Tree, Random Forest, and XGBoost algorithms are employed to create predictive models capable of distinguishing between legitimate and malicious URLs. The research involves the collection of a comprehensive dataset comprising both legitimate and various forms of malicious URLs. Feature engineering techniques are applied to extract relevant information from the URLs, transforming them into numerical representations suitable for machine learning. The three selected algorithms are individually trained and finetuned using the dataset, exploiting their unique capabilities to distinguish patterns indicative of phishing attempts and masked URLs. The performance of each model is evaluated using metrics such as accuracy, precision, recall, and web traffic. This study examines the application of machine learning algorithms to identify masked and phishing URLs. By comparing the results of these algorithms, a predictive model capable of distinguishing between legitimate and malicious URLs is created. Experimental results showed promising accuracy rates and potential to contribute to online security efforts. The implications of this research extend to advanced cybersecurity systems, offering enhanced protection against evolving threats in the digital domain.