Phishing URL Detection Using Machine Learning Techniques

Abstract

A criminal act performed online by impersonating others to obtain confidential data like passwords, banking details, login credentials, etc., is known as phishing. Detecting such websites in real-time, is a complex and dynamic problem, which involves too many factors. This work focuses on identifying the important features that distinguish between phishing URLs and legitimate URLs. To detect significant features, statistical analysis is done on the phishing as well as legitimate datasets. Based on the statistical exploration, certain features based on the URL, HTML, JavaScript and Domain were extracted. The prominent and most relevant features to identify the phishing URLs are identified using correlation. The identified subsets of features are then used to train different machine learning based classifiers and the accuracies obtained have been compared. From the experimental analysis it is observed that the extracted features have efficiently detected phishing URLs and the Decision Tree classifier has found with highest accuracy for making the predictions.