Matus UHLAR - fantomas skrev den 2014-07-13 12:33:
On 12.07.14 19:45, Benny Pedersen wrote:
meta BAYESSTOPPER (BAYES_99 && BAYES_999)
score BAYESSTOPPER 1.5 1.5 1.5 1.5
tflags BAYESSTOPPER noautolearn
This is useless... BAYES_99 applies whenever BAYES_999 does.
If you trust it, you can simply set BAYES_999 to 1.5 (or, add 1.5 to
current 0.2)
fair, rule should have being a softscore change like this:
score BAYES_999 (0) (0) (1.3) (1.3)
why this way ?, its counts on spamassassin corpus scoreing so it would
adjust depending on this
i hate hardscoreing localy since it would be my own fault then
tip perl evaluates scores in ()
