An Automated Framework for Real-time Phishing URL Detection

Abstract

An increasing number of services, including banking and social networking, are being integrated with world wide web in recent years. The crux of this increasing dependence on the internet is the rise of different kinds of cyberattacks on unsuspecting users. One such attack is phishing, which aims at stealing user information via deceptive websites. The primary defense against phishing consists of maintaining a black list of the phishing URLs. However, a black list approach is reactive and cannot defend against new phishing websites. For this reason, a number of research have been done on using machine learning techniques to detect previously unseen phishing URLs. While they show promising results, any such implementation is yet to be seen. This is because 1) little work has been done on developing a complete end-to-end framework for phishing URL detection 2) it is prohibitively slow to detect phishing URLs using machine learning algorithms. In this work we address these two issues by formulating a robust framework for fast and automated detection of phishing URLs. We have validated our framework with a real dataset achieving 87% accuracy in a real-time setup.

Topics & Concepts