Detection of Phishing URLs Using Heuristics-Based Approach

Abstract

Phishing is one of the types of cybercrime in which the attacker poses as a trustworthy entity with a view to obtaining sensitive information or data from the victim, this occurs usually through email. In the process, the victim may release information such as login credentials, credit card details, and other personally identifiable information that normally should not be revealed. The existing approaches used for phishing detection, therefore, need to be enhanced to effectively detect phishing. This study proposed a novel method for detecting phishing based on some heuristic features by extracting some relevant attributes, filtering these attributes, and classifying the same according to their impact on a website. The data explored for this study was retrieved from PhishTank and Alexa, which was later preprocessed for smooth model creation in python. The model created was evaluated and consistently gives a true positive rate of 85% based on the threshold set and an accuracy of 95.52%. The resulting output of this study has shown its reliability in the detection of phishing and could serve as a good benchmark for similar studies.