Monitoramento e identificação de páginas de Phishing

Abstract

Phishing campaigns frequently employ Web pages that mimic legit- imate pages to fool victims into providing sensitive information. Despite the research community’s continued effort to address these malicious activities, phishing becomes ever more sophisticated and continues making victims. In this paper we present a new phishing Web page monitoring framework that combines multiple techniques to achieve scalability and effectiveness. We also study trade-offs in the complex task of training models for identifying phishing Web pages. We show that representative datasets and features about the content of phishing Web pages are crucial for building general models. Our framework has been applied to hundreds of thousands of daily e-mails, and identified about one hundred phishing pages, a reduction of three orders of magnitude, and also serves as starting point for future phishing mitigation efforts