On Tuesday 01 June 2004 18:31, Spam Admin might have typed: > > What score does Bayes assign? > > Have you told Bayes that these mails are spam? > > It's doubtful I'm using Bayes, as this is simply a "filter and forward" > server to an internal GroupWise system.
Worth poking with a global bayes db. It's not as good as a per-user, but it still works pretty well. > There are URLs, none are consistent. Still worth trying the URI DNSBL patch for SA though, and enabling a few of the surbl.org checks. I run two or three of the surbl.org checks, and some days a single mail will hit all of them, racking up a large number of points instantly. > I forwarded a half-dozen of them to a person on this list, and he > pointd out that each body had a consistent mailing address in Toronto > (THUD! I missed that) so I will write a filter for that...however, I'm > still concerned that such explicit words are getting past these An explicit word, in and of itself means nothing. You could block explicit words (if they aren't graphics), but then you tread into the territory of IBM and the word 'die' :) Bayes helps a lot when dealing with content that static filters can't trap - worth giving it a spin. Most mails that come to my servers with the random 'poison' trigger a 5.4 (Bayes 1.0000) score simply because the combination of words is never seen in normal mail.
