Classification of Phishers

Abstract

Phishing attackers masquerade as genuine senders and try to steal consumers' personal identity data and financial account credentials. In spite of aggressive efforts, technology companies have had limited success in restricting phishing attacks. Unfortunately the nature of phishing attacks changed over time from passive, such as password guessing and eavesdropping to active attacks, such as employing Trojans to intercept traffic and adopting social engineering techniques. No matter how many authentication techniques we develop, phishers always adapt. However, phishers cannot become part of the recipient's social network without consent. Though they can forge certain fields in an email header, phishers do not have access to the complete header. In this paper, we describe techniques for detecting phishers based on their traffic paths, traffic patterns, and on the receivers' social network. Considering such issues, we based our solution on the trustworthiness of the relays participating in routing the emails. We examine the email's header rather than the content. We designed our classifier to perform the following analyses in four steps: i) DNS-header analysis, ii) Social network analysis, iii) Wantedness analysis, and iv) Proactive classification. We classify phishers into: i) Serial phishers, ii) Recent phishers, iii) Prospective phishers, and iv) Suspects. Finally, our classifier appends an alert level or label to the email's "subject" before adding the email to the inbox.