StackedPhish: A Stacked Ensemble Framework for Identification of Phishing Website
Mahbub Murshid, Mohammed Nasir Uddin, Fahima Hossain · 2023
The aim of phishing is to steal sensitive information such as usernames, passwords, social security numbers, credit card information, and so on from internet users. This research focuses primarily on presenting a strategy for detecting phishing websites based on machine learning. The approach takes into consideration lexical based features. Anti-phishing tools such as blacklists, whitelists, heuristics, and methods that are based on visual similarities are unable to detect newly launched websites. In addition, older methods are difficult to implement and are not appropriate for use in real-time contexts since they depend on the resources provided by a third party, such as a search engine. Therefore, one of the greatest challenges in the field of cybersecurity is the detection of recently constructed phishing web sites in an environment that operates in real-time. This work presents a lexical feature-based anti-phishing technique, which derives anti-phishing features from lexical features, as a solution to the challenges described above. The results of our experiment show that the phishing detection strategy StackedPhish using the Stacking Classifier is more effective, with higher detection accuracy of 99.39% for multi-class classification and 99.88% for binary -class classification.