New Research from the Proofpoint Attack Response Center

Exploiting Spammers' Tactics of Obfuscations for Better Spam Filtering

This research paper by Proofpoint Attack Response Center scientists Vipul Sharma and Steve Lewis explores how the text-based obfuscation techniques used by spammers can be detected, countered and even used as a reliable indicator of spam. It was originally presented at the Virus Bulletin 2006 conference, held in October 2006.

Please enter your personal contact information to download a free copy of this paper. All fields are required so your request can be promptly processed. To view our Privacy Policy, click here.

First Name:
 		Last Name:
Company:
 		Job Title:
Email:
 		Phone:
State/Province:
Zip/Postal Code:
Country:

Company Information

Please provide your company information (number of email inboxes and industry).
Company Size (# Mailboxes):
 		Industry:

Optional Information

Please let us know more about your email security and data loss prevention needs (not required).
I'm evaluating email security or DLP solutions:
 		Please have a representative call me now:

Abstract

Text-based obfuscation tactics are commonly used by spammers to fool content-based spam filters. The current methods used to counter such obfuscation are (1) de-obfuscating spam emails and (2) using obfuscation as a feature for spam classification. This paper discusses de-obfuscation methods that better fit the needs of enterprise-class spam filtering; presents an empirical comparison of various multivariate classification techniques for addressing the problem of obfuscation detection; describes a novel obfuscation detection model and the benefits of using auxiliary classifiers for better classification accuracy; and presents conclusions and a discussion of the overall study.