Explaining URL Phishing Detection by Glass Box Models
Sangeeta Mittal · 2023
Phishing is one of the successful cyberattack that targets unsuspecting users to make them divulge personal sensitive information. Uniform Resource Locator (URL) phishing particularly involves distributing address strings of phishing websites via various vectors like email, websites and phone SMSes. Feature engineering of these URLs reveals that extracting meaningful statistical information and training Machine Learning (ML)based models for identifying whether a URL is legitimate is an effective automated means of protecting users from these attacks. In this work, a case of Explainable AI (XAI) has been made to provide insights into how algorithms make predictions, enabling humans to understand and evaluate why the URL under question is not secure to visit. Various examples have been illustrated to visualize the explanations. The work can act as a basis for generating counterfactuals to minimize the unintentional composition of a URL as phishing.