DETECTION OF PHISHING UNIFORM RESOURCE LOCATOR (URL) USING SUPERVISED MACHINE LEARNING ALGORITHM

ABSTRACT:

The growth of the Internet and computer network security have aroused users' responsiveness to the secured network environment as this created a basis for the rapid and comprehensive progress observed in Internet patronage. Part of the challenges in the rapid internet growth is the internet security. Using phishing as an example in today's world of technology, can be considered a menace rising fast at the same pace technology is developing. This act may be viewed as an unlawful performance that mixes social-engineering approaches to deceive internet users and capture sensitive information. Phishing is known as cybercrime in which victims conversed through email, telephone, or text message by an attacker in the pretense of trustworthiness. Internet criminal frequently changes methods of perpetuating the acts of phishing such that traditional ways of identifying new phishing links may not be adequate again. It is therefore important to resolve to the predictive mechanism of ascertaining a newly emerging phishing website and rapidly improve the accuracy of the prediction. This paper considers a state-of-the-art survey on systems for website phishing detection and analyzes an approach that extracts features from phishing URLs and trains classification models with extracted features to detect phishing. Four Supervised Machine Learning algorithms — Decision Tree, Logistic Regression, Naïve Bayes, and Random Forest — were used as models. The dataset was derived generically from the field through selected online users for the purpose of this study. The four (4) algorithms were evaluated using the WEKA application, among the four Algorithms Logistic Regression proved to perform better in the classification with 99.88% Accuracy for the Phishing URL detections.

Keywords—Classification; Phishing; Phishing website detection; Machine learning