On Thu, Jun 17, 2004 at 08:25:07AM +0100, Duncan Hill wrote: > No Bayes? A well trained bayes will hit 5.4 points by default (99 - 100% > score), which would easily tip that mail from ham to spam. >
Unfortunately after all this time running my own mail server and using spamassassin, I'm just now getting around to playing with the bayesian filtering. I'm well on my way to training my system with spam, but most of the ham I get is mailing list mail; not much actual personal mail really. Are there some best practices for training the bayes system that I should keep in mind? Thanks! /Owen
