Phishing Website Detection Using Machine Learning

Phishing attacks have been one of the biggest cybersecurity concerns, where attackers develop fraudulent websites that resemble legitimate online services to steal critical user information such as login credentials, bank account information, and other sensitive information. The conventional anti-phishing detection systems, such as blacklist-based systems, rule-based systems, etc., are found to be less effective in handling the detection of new phishing sites that emerge frequently in the world wide web. To overcome the limitations of the conventional systems, the authors of the current research propose a supervised machine learning-based anti-phishing detection system using a set of discriminative features extracted from the address bar, domain-based features, and other components of the webpage such as HTML and JavaScript code. A set of machine learning and deep learning-based classifiers have been trained and tested using a balanced set of legitimate and phishing URLs to evaluate the performance of the proposed system. The experimental results show that the proposed system achieves better performance in terms of accuracy, precision, and recall using ensemble-based models such as XGBoost in comparison to other state-of-the-art models.