Howdy all - Currently we use procmail recipes to put tagged spam into an IMAP spam folder for each user. We also have a spam-to-learn folder for each user to put false negatives in. We run 'sa-learn --spam --mbox' against the spam and spam-to-learn folders for each user, and 'sa-learn --ham --mbox' against all their other folders via cron.
I use spamass-milter and am seriously considering bouncing all e-mails with an SA score of 10 or more. I personally am getting over 200 spams a day. If I bounce everything over a 10, I will reduce that by over half. By cutting the sample of spams that sa-learn can use, how will I affect the Bayes database? Since the spams in my spam mailbox will have a lower average score than before, will that hurt or help Bayesian learning? Thanks! -- Thomas Cameron, RHCE, CNE, MCSE, MCT Cameron Technical Services, Inc. (512) 454-3200 Main http://www.camerontech.com ------------------------------------------------------- This SF.Net email sponsored by: Free pre-built ASP.NET sites including Data Reports, E-commerce, Portals, and Forums are available now. Download today and enter to win an XBOX or Visual Studio .NET. http://aspnet.click-url.com/go/psa00100006ave/direct;at.asp_061203_01/01 _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
