bogofilter producing poor results

[Greg Louis]

On 20021112 (Tue) at 0727:50 -0500, Greg Louis wrote:
> for Robinson you
> need to tune SPAM_CUTOFF and perhaps ROBS and ROBX (the default value
> of 0.2 for the latter is probably a bit low, I use 0.415)

I should have added: ROBX is important when the training set is small,
because in that case there will be more unknown words in test messages. 
ROBS will determine how big the training set has to be before ROBX
dwindles in importance; it's a kind of strength indicator for the trust
we put in ROBX, as Gary points out in his essay.

-- 
| G r e g  L o u i s          | gpg public key:      |
|   http://www.bgl.nu/~glouis |   finger greg at bgl.nu |