From: "John Broadhead" <[EMAIL PROTECTED]>
> I have spamassassin installed on our mail server. On my account it seems
to be
> working fine, messages often get thrown in my Junk folder based on my
rules. If
> I look at them they will always have a Bayesian score (Very often 1.0000).
>
> My friend's account is on the same system and he also has spamassassin set
up
> to filter email. He gets a lot more spam and puts it in a special folder,
which
> he daily runs "sa-learn --spam --mbox" on. The weird thing is that
spamassassin
> never seems to attach a bayesian score to the email.
>
> I've looked at the headers from both his spam and his ham and his
misclassified
> ham, and none of them seem to ever have a bayesian score of any type, not
even
> BAYES_50 or something neutral.
>
> Is there anything I'm missing to "turn on" bayesian filtering?
Training on spam is useless without concurrently training with ham.
Otherwise the Bayes filter has nothing to use for discrimination.
{^_^}
