Vee Persaud said: > I received an email on one of my systems that is not running SA and I > decided to forward it to my SA system to see if it would get rejected. > The subject was "Have amazing s\ex up to 20 times per day lugpsxw". > They use the "/" in the body also to avoid spam detection. It got a score > of 3.3. I ran sa-learn --spam on this message and now I get a score of > 3.0 if I resend the message. I'm a bit confused. Did I do something > wrong ???
Vee, Please send the headers from both the first and the second messages. Without more information all anyone can do is wildly guess. Also, it's worth noting that learning a message as spam does not necessarily mean it will be marked as spam if tested again. Instead of learning the entire email as a spam (which would be very rigid), the tokens inside the spam are added to the bayes db, which is an aggregate of spam AND ham tokens. The next time you check another copy of that spam, the tokens in the spam are checked against the tokens in the bayes db and are statistically analyzed to create a spam probability. This probability is used to compute a score that is added to the spam core. ------------------------------------------------------- This SF.net email is sponsored by OSDN's Audience Survey. Help shape OSDN's sites and tell us what you think. Take this five minute survey and you could win a $250 Gift Certificate. http://www.wrgsurveys.com/2003/osdntech03.php?site=8 _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
