On Tue, Apr 15, 2003 at 03:41:24PM -0700, Dennis Gearon wrote:
> What would be a more friendly blocking message would be an invitation to be
> on the white list. I've seen a very slowly increasing trend for that. My
> prognostication talent says this is the wave of the future.
I say stop using blacklists and start using bayesian spam classifiers
(recent SpamAssasin for example). It's more costly because you have to
filter mail coming from known spammers, but you get much less false
positives. You also have to train it.
I actually filter all my pgsql lists, plus some spanish lists, through
one we made, and get 5-6 false negatives a day (spam as non-spam), and
3-4 false positives a week (non-spam as spam). 30-40 spam messages are
correctly detected daily. I dunno how much "real" email, but around 100
or 120 I think. With a well-written classifier the scores are better;
this one is more of an experiment, and it rarely gets feedback (good
ones feed every message into scorefiles).
--
Alvaro Herrera (<alvherre[a]dcc.uchile.cl>)
La web junta la gente porque no importa que clase de mutante sexual seas,
tienes millones de posibles parejas. Pon "buscar gente que tengan sexo con
ciervos incendiánse", y el computador dirá "especifique el tipo de ciervo"
(Jason Alexander)
