Bayes has been working great. We use it site-wide here, with corpora of about 20K spams and hams. We used to get anywhere from 4-12 false positive reports per month, requiring whitelisting (we have about 6K email users). Now we get none, and I attribute that to Bayes. Nice job, SpamAss authors, with implementing this!
However it seems the latest rounds of Nigerian Scam mails look like ham to Bayes. All of them are scoring bayes_00, which brings them below our threshold. I keep feeding copies of it into sa-learn, but without luck. Anyone have some custom rules that would help here? I want to keep our 0 false positives record, so I am hesitant of just increasing the NIGERIAN rules by 4 or 5 points to coutner the bayes_00 negative score. Thanks, Matt -- Matthew S. Cramer <[EMAIL PROTECTED]> Office: 717-396-5032 Infrastructure Security Analyst Fax: 717-396-5590 Armstrong World Industries, Inc. Cell: 717-917-7099 ------------------------------------------------------- This SF.net email is sponsored by: SF.net Giveback Program. Does SourceForge.net help you be more productive? Does it help you create better code? SHARE THE LOVE, and help us help YOU! Click Here: http://sourceforge.net/donate/ _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
