On Thu, Jul 10, 2003 at 11:53:01AM -0300, German Staltari wrote: > Hi, i've attahced a very tricky SPAM mail that has been scored with 0.8 > points, what can be done with this kind of SPAM?.
First and foremost, you can feed it to the Bayesian classifier as spam using sa-learn. That will cause all the "interesting" and obfuscatory misspellings to be recognized and stored as tokens for use in scanning later mail. This is, of course, yet another arms race. I've recently seen spam consists of nothing but attachments, so that they show up empty when using mutt to read them, until one types "v" to see all the attachments. From time to time, one of the mroe enterprising spammers sends mail with nothing *but* images, so that I either have to classify as spam all such mail or to run OCR software on the images and pass the resultant text through SA. -- Mike Andrews [EMAIL PROTECTED] Tired old sysadmin since 1964 ------------------------------------------------------- This SF.Net email sponsored by: Parasoft Error proof Web apps, automate testing & more. Download & eval WebKing and get a free book. www.parasoft.com/bulletproofapps _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
