Benchmarking Machine Learning Techniques for Phishing Detection and Secure URL Classification

Abstract

Phishing is still one of the biggest threats in cybersecurity. It is the exploitation of users through the use of deceptive URLs. In this study, the outcomes of the Random Forest, Support Vector Machines, and Decision Tree models are analysed on a dataset containing more than 640,000 URLs. The results showed that Random Forest recorded the highest accuracy of 87.85% on the Aalto dataset and 86.86% on the Kaggle dataset. These perspectives provide a more fact-based approach towards developing more effective and practical anti-phishing systems.

Topics & Concepts