Detection of Phishing Website Using Machine Learning and Features Extraction

Computer security

Phishing assaults are fast-expanding menace in cyberspace which costs web users and organizations huge financial loss annually. Sensitive information obtained from customers via various social engineering methods is prohibited, including Web sites, pop-up messages, instant messaging, email, and other communication channels can all be utilized to spot phishing attempts. This paper provides a framework that can identify phishing or real URL links. A collection of harmless, junk mail, malware, phishing, and defacement URLs are included in the data set employed for categorization. Additionally, phishing URLs from an open-source platform called "Phish Tank," which offers phishing URLs in various forms like JSON, CSV, and others, are included. Six (6) models of machine learning and deep neural network techniques are used to identify phishing URLs. With a collection of over 10,000 randomly chosen URLs, split into 60% training and 40% testing samples, and comprising up to 23,328 phishing and 4894 valid URLs, the research purpose is the development of online applications that can quickly recognize phishing URLs. The Uniform Resource Locator datasets has been trained and evaluated utilizing feature selections such as HTTPS & JavaScript-based features, domain-based features, address bar-based features in order to differentiate among legitimate and phishing URLs.  The research provided a method for classifying URLs into legitimate and fraudulent URLs. In order to help individuals and organizations spot phishing links and stay one step ahead of the criminal, it would be very beneficial to authenticate each link that is delivered to them in order to verify its credibility.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.