Phishing Diagnosis: A Multi-Feature Decision Tree-based Method

Abstract

Phishing is an electronically connected criminal activity in which the attacker steals the user’s personal information like username, countersign, internet banking account, credit/debit card number with the expiration date, password, pin, legitimacy, confidential patient record, CVV number, etc. to boon financially. Email-based phishing is the most common and traditional way of phishing scams, in which the phisher will send a suspicious email with an embedded URL and ask the user to click the URL. When the user clicks on the link, the link will be redirected to a spoofed site that looks the same to the original site to steal their credentials and displays some error message. Later the phishing uses those credentials for malicious purposes. To overcome these scams, many anti-phishing tools have developed. Among that the machine learning-based approaches can give a better result. This paper is an extensive study of the various machine learning-based anti-phishing approaches and their results that detect the phishing URL's from the URLs with URLs features. Six most important models of machine learning have been examined for the phishing detection problem. The Decision Tree-based method outperforms other methods.