Ah, good. Your other option is to go get a bunch of extra rulesets, from:
http://www.rulesemporium.com/ and/or http://www.exit0.us/
I personally use:
70_sare_bayes_poison_nxm.cf
99_sare_adult.cf Obfu1.cf bigevil.cf header_abuse.cf myrules.cf (this is my own collection, obviously)
virusbounce.cf
70_sare_random.cf 99_sare_spoof.cf antidrug.cf coding_html.cf ratware.cf
Steve Mansfield wrote:
------------------- On Friday 14 May 2004 14:14, Kevin Peuhkurinen wrote: | You could try lowering the spam threshold if you see that your ham is | getting scored low enough to avoid becoming FP's. I found that as | Bayes got better, I was able to lower the threshold by 0.5 from my | starting point of 9 every week or so until I reached my current | threshold of 3.5.
I have, in fact, started down that road! My site-wide setting is 10 - ie, lets loads of stuff through and I've been reducing my user settings - currently at 4 and still not getting any false-positives. But I've seen some spam coming through with ratings much lower - eg, a message entitled
"9-need a loan8 it takes 2 minutes to apply" got a rating of:
X-Spam-Status: No, hits=1.7 required=4.0 tests=HTML_30_40,HTML_MESSAGE, MIME_HTML_ONLY autolearn=no version=2.63
I've set up filters in Kmail using regular expressions to weed out mails offering certain performance-enhancing drugs, and as that filter comes before the SA one for emails coming direct into Kmail from the POP server (rather than the Fetchmail route) then I'm already weeding out a lot of stuff that way.
