Performance study of classification techniques for phishing URL detection

In the constantly evolving World Wide Web environment, the end-users are subjected to many threats. Phishing is the most relentless of all the attacks. Detection of phishing URLs from genuine ones is a paramount task to minimize the financial loss incurred. By applying pattern recognition capabilities of machine learning to phishing detection domain, we can achieve significant performance improvements. This paper provides a survey of research works conducted on classification techniques by various researchers for phishing URL detection. The experiments were performed using 4,500 URLs and several classification algorithms. The observed results showed that tree-based classifiers provide maximum accuracy.