[Spambayes] General training questions
Hans Henderson
hans at pobox.com
Tue Oct 12 02:48:34 CEST 2004
Sorry to re-post if it's been answered elsewhere, but I believe others have
also asked similar questions I didn't see answered.
So far my intention is to put all "unsure" and all wrongly classified mail
into the training folders.
The docs state to try to keep the quantity of spam and ham trained
balanced, but I get a LOT more spam than ham (like 2000:1).
I've done this with the initial training set, but unfortunately my ham
sample was much older than the spam, as I don't keep old spam <g>
But ongoing, is there any harm in the ratio in the database getting
unbalanced from putting a larger amount of spam than ham from the unsure
folder into the training folders?
Or is it better to deny SB the training fodder?
More information about the Spambayes
mailing list