Machine Learning based URL Analysis for Phishing Detection

Attacks that take place online most frequently take the form of phishing scams. Attackers attempt to gather user data without the consent of any such user through emails, URLs, and any other link that sends a viewer to the a dubious page where a consumer is persuaded to start taking specific actions that can successfully complete an attack. An attacker has the opportunity to gather vital information about the victim during these attacks, which they can use to presume the victim’s identity & carry out tasks that only the victim should have been able to carry out, such as making purchases, sending messages to the other people, and simply attempting to access the victim’s info. Many studies have been done to discuss potential defenses against these assaults. This study employs three algorithms for machine learning to ascertain whether such a web page is phishing. In the experiment, software that analyzes web page URLs to distinguish between legitimate & phishing websites is used to try to prevent attacks using these models which have been trained to utilize URL-based features. The accuracy, recall, and F1 Score of the random forest classifier’s performance from the observations were 97.5%, 99.1%, and 97.3%, respectively. The proposed model is quick and effective because, unlike earlier studies, it only relies on the URL and doesn’t conduct analysis using any other sources.