Phishing Detection Using URL-based XAI Techniques

Abstract

The Internet has been growing exponentially and expanding facilities, such as payments and online purchases. Likewise, the number of criminals using electronic devices to commit theft or hijacking of information has increased. Many scams still require interaction with the victim, who in many cases is persuaded to access a malicious link sent by email, which is classified as phishing. This technique is one of the biggest threats for users and one of the most efficient for criminals. Several studies show different sorts of protection using Artificial Intelligence, which despite being very efficient, do not describe the reasons for categorizing them or using a URL as phishing. This paper focuses on detecting phishing using explainable techniques, i.e., Local Interpretable Model-Agnostic Explanations and Explainable Boosting Machine, to lighten up new advances and future works in the area.