RE: reliable email

  • Authors:
  • Scott Garriss;Michael Kaminsky;Michael J. Freedman;Brad Karp;David Mazières;Haifeng Yu

  • Affiliations:
  • Carnegie Mellon University;Intel Research Pittsburgh;New York University and Stanford University;University College London;Stanford University;Intel Research Pittsburgh and Carnegie Mellon University

  • Venue:
  • NSDI'06 Proceedings of the 3rd conference on Networked Systems Design & Implementation - Volume 3
  • Year:
  • 2006

Quantified Score

Hi-index 0.00

Visualization

Abstract

The explosive growth in unwanted email has prompted the development of techniques for the rejection of email, intended to shield recipients from the onerous task of identifying the legitimate email in their inboxes amid a sea of spam. Unfortunately, widely used content-based filtering systems have converted the spam problem into a false positive one: email has become unreliable. Email acceptance techniques complement rejection ones; they can help prevent false positives by filing email into a user's inbox before it is considered for rejection. Whitelisting, whereby recipients accept email from some set of authorized senders, is one such acceptance technique. We present Reliable Email (RE:), a new whitelisting system that incurs zero false positives among socially connected users. Unlike previous whitelisting systems, which require that whitelists be populated manually, RE: exploits friend-of-friend relationships among email correspondents to populate whitelists automatically. To do so, RE: permits an email's recipient to discover whether other email users have whitelisted the email's sender, while preserving the privacy of users' email contacts with cryptographic private matching techniques. Using real email traces from two sites, we demonstrate that RE: renders a significant fraction of received email reliable. Our evaluation also shows that RE: can prevent up to 88% of the false positives incurred by a widely deployed email rejection system, at modest computational cost.