On Thu, 31 Oct 2002, Barry A. Warsaw wrote:
>
> People who are interested in better spam blockers should check
> out the spambayes project on SourceForge:
>
> http://spambayes.sf.net/
>
> This is a Bayesian classifier for detecting spam. When properly
> trained (and that's the trick) it has extremely low false
> positive and false negative rates.
Steve Gibson and another guy are working on a variant of
Bayes which self-trains; more at grc.com, and especially on
spamnews at news.grc.com. I asked yesterday if/when it'll run on
linux; haven't checked yet.
And just to jump *way* out of my depth: I seem to recall
that the excellent AdSubtract is starting to get into anti-spam. If
that's feasible for it, it may also be feasible for Privoxy!
--
RR 'Beartooth' Neuswanger <karhunhammas (at) lserv.com>
On the Internet, you can never tell who is a dog --
supposing you care -- but you can tell who has a mind.
