Spam control

1
Spam control

• Old emphasis detect spam
• black lists, Bayesian content analysis
• danger discard non-spam (false positives)
• often easily subverted by spammers
• CUIT measurements most SpamAssassin filters are
  never triggered
• Also now needed for VoIP (SPIT)
• content-based techniques are not applicable

2
Spam phishing prevention

• Our emphasis detect ham (non-spam)
• not content-based
• Approaches
• HTTP extensions for transactional email (Amazon,
  airlines, ...) and phone calls
• add response header with related mailto and SIP
  URLs
• social and professional networks
• e.g., outbound email and calls
• IETF mailing list archives
• EDAS (most networking researchers)
• IEEE/ACM member lists
• vast majority of ham is from users or domains
  known to receiver
• domain age (whois)
• URL lifetime for phishing detection