[Spambayes] Is Equal Ham & Spam really the best?

Amedee Van Gasse amedee at amedee.be
Mon Jul 30 23:23:34 CEST 2007


Op maandag 30-07-2007 om 09:48 uur [tijdzone -0500], schreef
skip at pobox.com: 
> Amedee> I'm using the procmail filter.
> 
> You're using "sb_filter.py -t" (or sb_bnfilter.py)?

I'm using sb_filter.py to score messages.
For training, I use
sb_mboxtrain.py -n -r -d $HOME/.hammiedb -g
$HOME/Maildir/.ztrain.confirmed-ham -s
$HOME/Maildir/.ztrain.confirmed-spam

> I'd recommend that you also save the messages you train on

OK.

> and occasionally retrain from scratch if you discover you've made a
> mistake.

So far I have not yet made a mistake, because when I move mail to a
traing folder, it only gets trained on the next daily cron run. That
means I have enough time to move misclassified mail.

> You might also try the train-to-exhaustion script in contrib/tte.py 

OK. How does that work? With sb_mboxtrain.py I just "drag & forget". Can
I do the same with contrib/tte.py? Can I use it as a drop-in replacement
and let a cron job run on it every night?

> and only use the sb_filter script to score messages.

OK, I'm already doing that.

-- 
Amedee Van Gasse <amedee at amedee.be>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Dit berichtdeel is digitaal ondertekend
Url : http://mail.python.org/pipermail/spambayes/attachments/20070730/c0d7ba82/attachment-0001.pgp 


More information about the SpamBayes mailing list