Due to the nature our setup here I have 250 users that receive email 
forwarded from other accounts on external mail servers.  The only 
anti-spam processing I can apply to these emails here is content based. 
  Whilst using URIBL and Clamd with Sane Security sigs catches a lot of 
SPAM it falls upon bayesian analysis to catch the majority.

I'm really struggling to get bayesian analysis working good enough to 
keep the users happy though.  I get a good number of emails that are 
rejected that are legitimate and if I turn on confidence threshold I get 
a lot of spam passed through because the confidence level is 0.

Could anyone provide any pointers that might help me fine tune my setup 
and get it working better ?  I'm guessing it's probably just a case of 
forwarding wrongly identified emails back to the email interface to 
train it - my users simply won't do this though and I have to do it 
myself :|

Any help would be appreciated :)


-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
Assp-user mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/assp-user

Reply via email to