On Sun, 2004-05-30 at 11:55, Hannes Mayer wrote: > Well, teaching it the spam that gets thru is difficult in my case, cause > there are many users and they download the messages via POP3. > Identified spam goes into a seperate folder (and I rotate those folders > a few times until they get deleted - like logrotate), so feeding those > messages is no problem. > Do you have POP3 users on your server ? How do you feed unflagged spam ? > > Thanks a lot, > Hannes. I setup a spamuser that missed messages can be bounced to. (NOTE: you want to use your clients bounce option not forward. Forwarding a message changes the headers by putting your address in as the from. If you teach it that your address generates spam you will have more problems. By bouncing the message the headers should be retained and you can teach spamassassin to correctly identify those as spam.) In my case I also divert any spam to that dummy user via procmail. This allows someone to review the tagged messages just in case there was a false positive. I read some notes on the spamassassin.org site about fully automating this process but I have not done that. In theory you could have an account that when a message is sent to it you would run sa-learn on the message automatically. You would have one for spam and one for ham. That way your end users could send missed spam and have it added to the database as well as send a ham message that might have been flagged incorrectly. I did not trust that process nor do I trust users that far. <GRIN> We use pop3 for our users. All in all it has worked far above my expectations. If everyone used it the spammers would be out of buisness IMHO. -- Scot L. Harris <webid@xxxxxxxxxx>
