Detection of Fake Websites using Machine Learning Techniques

Abstract

Phishing websites are harmful websites that spoof legitimate web pages to get sensitive information from users as login, account, and bank card info. Detecting these hoax websites is a difficult topic since hacking is mostly a semantics-based assault that targets human vulnerabilities rather than network or system flaws. Machine learning systems can identify phishing assaults and have greater adaptability for forms of hack attempts, hence these are widely used. To employ this sort of method, input characteristics should be properly chosen. These aspects determine the overall performance of the solution. In this paper, two techniques Logistic Regression and Multinomial Naïve Bayes are extensively used in detecting these websites using phishing-url datasets. Out of these, Logistic Regression has achieved the highest accuracy results of 97%.