Anti-Spam Techniques

Anti-spam techniques table

August 29, 2003

Anti-Spam Techniques
TECHNIQUE	PROS	CONS
Keyword Filtering – Specific strings are searched for in an E-mail message.	Simple to implement. Already available in many E-mail packages.	Not very accurate. Frequent false positives (good E-mail messages flagged as spam)
Rule-Based Filtering – E-mail messages are filtered and scored by keyword and context using a set of rules.	Can be very accurate.	Requires ongoing rule updating. May miss well-crafted spam messages.
Bayesian Filtering – A type of statistical analysis derived from Bayesian logic.	Learns to recognize spam. Returns percentage probabilities instead of scores. Can be very accurate.	Depends somewhat on the quality of spam used to teach it. May need to be customized for each user.
Blacklists – Lists of IP addresses that have been identified to originate spam.	Good at blocking known spam sources.	Requires constant updating. Misses well-crafted spam messages and spam that isn't coming from known mail relays.
Fingerprinting – Identification of spam based on similarity to previously received spam messages.	Can quickly recognize spam with a high rate of accuracy.	Doesn't recognize new spam messages. Requires constant updating from a central source.
Challenge-Response – E-mail senders are challenged to prove their identity before E-mail messages are delivered.	Very high degree of spam prevention.	Creates more E-mail traffic. May block valid E-mail sent by machines. May require foreign-language support.
Secure Messaging – E-mail is sent with encrypted credentials or on an encrypted transport, so that the sender or sending machine can be authenticated.	Very high degree of spam prevention. Reliable identification of E-mail source.	Requires widespread acceptance and support for digital-identity standards. May require expensive infrastructure enhancements.