Hello Scott, John, Wednesday, January 21, 2004, 9:07:01 PM, you wrote:
SW> John, SW> After this I started adding custom cf's like Bigevil,backhair, SW> etc. I also started pushing some of the default scores up and down SW> according to what my beta testers were seeing. I'm now starting to SW> back off on the threshold to eventualy get it back to 5.0. CF's like SW> bigevil are expecting your threshold to be 5.0. I believe even Bayes SW> assumes your threshold is 5.0. ALL the standard scores in the distribution set assume a threshold of 5.0 If you're running with a low threshold, probably anything under 4.0, you should review the scores in 50_scores.cf, and maybe reduce some of the higher scoring rules to make sure that no one or two rules can by themselves flag an email as spam. IMO you generally want three or more rules to kick in before an email is flagged as spam. (There are exceptions, and I do have a small number of rules that by themselves kick in the spam flag, such as emails sent to [EMAIL PROTECTED] ...) Likewise, those who run with increased thresholds like my 9.0 often need to (cautiously) increase the scores of some rules, so they don't simply lose effect. I've increased the scores of several dozen rules (and even with my 9.0, decreased the scores of a handful). This customization is part of the power of SA -- your spam differs from my spam, and so our rules should differ also. The distribution scores are an excellent starting point (most scores are not changed), but with a little due diligence and caution, they can be improved upon. Bob Menschel ------------------------------------------------------- The SF.Net email is sponsored by EclipseCon 2004 Premiere Conference on Open Tools Development and Integration See the breadth of Eclipse activity. February 3-5 in Anaheim, CA. http://www.eclipsecon.org/osdn _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
