Gary, doesn't that presuppose that the mail/lists directory does not
contain a spam list?
I also sense a lack of spam training here. One sided Bayes training is
not a good thing.
{^_^}
From: "Gary D. Margiotta" <[EMAIL PROTECTED]>
#!/bin/sh
cd mail/Lists
for x in `ls`
do
sa-learn --ham --mbox $x
done
-Gary
On Fri, 2 Jun 2006, Kenneth Porter wrote:
On Friday, June 02, 2006 9:47 PM -0400 JamesDR <[EMAIL PROTECTED]>
wrote:
How many messages have you trained? You'll need 200 each to get it going,
and I recommend at least a thousand of each to really get it going.
I use procmail to distribute my mail to over a hundred folders in a large
tree, mainly to deal with mailing lists and to separate mail from friends and
coworkers. Has anyone come up with good tools for dealing with a hierarchy of
mbox files when using SA? For instance, it would be convenient to have
sa-learn start at the top of my mail/Lists hiearchy for ham training. I'd
also like to run mass-checks against my hierarchy.
