Abstract

The Internet serves as a global crossroads, enabling users to connect and share information. Unfortunately, this connectivity has become a breeding ground for malicious activities, with phishing attacks emerging as a significant threat. Phishing attacks represent a severe danger to online accounts and data due to their deceptive nature. The average person now finds it increasingly challenging to differentiate between a legitimate email, link, or website and a fraudulent one. It is against this backdrop that the research is initiated. The research paper introduces an effective machine learning algorithm designed for the identification of both phishing and legitimate URLs. By integrating logistic regression and pipeline, the algorithm is adept at processing a sizable dataset comprising 549,347 legitimate and phishing URLs. Notably, the experimental results underscore the algorithm's superior accuracy in distinguishing phishing URLs compared to previous research efforts. This emphasis on leveraging a substantial dataset and achieving enhanced accuracy sets the proposed technique apart in the field.