FRAMINGHAM (09/22/2003) - Network World's anti-spam fighter, Peter Hebenstreit, recommends familiarity with these methods of stopping unwanted e-mail.
Attachment checking - Checks for macros and text in attachments.
Blacklists - Let users designate a source or IP address from which no mail will be accepted.
Code checking - Looks for "open new window" or any other type of scripting that might be malicious.
Complex dictionary checking - Screens text for no-no words and won't be fooled by various tricks, such as the replacement of letters with look-alike numerals (c001 = cool).
Content checking - Scans plain text for key phrases and the percent of HTML, images and other indications that the message is spam.
Header checking - Checks for valid Multi-purpose Internet Mail Extensions content, valid Simple Mail Transfer Protocol addresses, and
Heuristics - A checklist of rules and tests to determine mathematically the likelihood that the message is spam.
Machine-learning or self-learning - Methods, such as Bayesian filtering, by which the filter can create and update its own heuristics checklist.
Snapshotting or fingerprinting - Identifies that similar, yet not identical, messages are part of the same, already-identified spam broadcast.
Whitelists - Lets users designate a source or IP address from which all mail will be accepted, even if individual messages earn high spam ratings.
- Julie Bort