EXPLORING A ROBUST MACHINE LEARNING CLASSIFIER FOR DETECTING PHISHING DOMAINS USING SSL CERTIFICATES

Abstract

Due to the HTTPS phishing sites appearing genuine to users, detecting them is a challenging task. In this thesis, I have studied the important attributes of how attackers use SSL certificates in sites with fake domains. The robustness of a system is explored to auto-detect a phishing site using the critical attributes of an SSL certificate. Considering good performance and transparency in the resulting model, I have chosen the decision tree algorithm for decision making of site category. The proposed classifier has achieved around 97\% of correctly classified instances in comparison with other machine learning classifiers. In order to connect users to the system, a Web API is created which provides the user interface of the proposed system through HTTP service. Evaluation results show the promising effectiveness and efficiency of the Web API system designed and developed.