Jack Gostl writes: > That new Bayes algorithm is mighty touchy. So far its tagged four real > messages with a BAYES_99, three of them today alone. In just five days it > has had twice the false positives that 2.55 had in four months. > > That's a bit sensitive for my blood. I'm not sure I can release this to > my user community with that kind of sensitivity. My inclination is to > reduce the score on BAYES_99, but I'm open to suggestion.
Have you tried checking to see what tokens it's picking up, and what they look like in the db? 1. get the message 2. run "spamassassin -D -t < msg > out", and look for the Bayes tokens listing in the STDERR output 3. examine that, and compare with output from "sa-learn --dump" to see what the token probabilities look like. It could be that some bad training data has crept in somehow into the db -- and that's why Bayes is FPing. --j. ------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
