Folks,

We're having a lot of trouble with the bayes module. It's only learning spam messages, and is now(despite having been trained on thousands of "ham" messages) marking virtually anything, even messages with just "test" for a subject/body, as spam- 99% certainty, no less.

I tried setting bayes_auto_learn_threshold_nonspam to a positive value- almost all legitimate email we get on the particular system is marked somewhere between 0 and 2- never any lower. Ever(save for whitelisting).

That setting appeared to have no effect. I just sent a test, it got a score of 0, and was tagged autolearn=no. Even after I removed the whitelists. I need to change this, quickly.

Also- maybe it's just me, but it seems rather silly to not allow the user to auto-learn messages that have been whitelisted, either sitewide or user-specific. Could someone a)explain the reasoning here and b)tell me how to change this? It is almost completely contrary to what we want- any addresses we've whitelisted are guaranteed to be sending legitimate email and we would ABSOLUTELY want them auto-learned as ham, NOT the other way around!

Frankly, SA seems to have a lot of little eccentric settings you can't override even from the config file and it's my biggest complaint about the package. I question the wisdom of "the user don't know better" type of programming and silently overriding what the user has asked for...or making policy decisions and not allowing any way to override(ie, the never-learn-whitelisted-mail).

Brett



-------------------------------------------------------
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

Reply via email to