Consider doing training via redirects or IMAP.
http://wiki.apache.org/spamassassin/ProcmailToForwardMail
http://wiki.apache.org/spamassassin/RemoteImapFolder
- dan
--
Dan Kohn <mailto:[EMAIL PROTECTED]>
<http://www.dankohn.com/> <tel:+1-650-327-2600>
-----Original Message-----
From: Tony Finch [mailto:[EMAIL PROTECTED] On Behalf Of Tony Finch
Sent: Thursday, July 15, 2004 07:02
To: Dan Kohn
Cc: [EMAIL PROTECTED]
Subject: RE: paper comparing spam classifiers
On Thu, 15 Jul 2004, Dan Kohn wrote:
> http://plg.uwaterloo.ca/~gvcormac/spamcormack1.pdf
Thank you very much.
This paper also points out that SpamAssassin in auto-learn mode (i.e.
without any manual training) is significantly better with both
false-positives and false-negatives than SpamAssassin without Bayes.
This
is important for sites like mine where the installation is centralized
and
there is no sensible way of getting training data. I'm currently running
without Bayes for this reason, and because of vague FUD, and probably
also
hardware performance limitations.
Tony.
--
f.a.n.finch <[EMAIL PROTECTED]> http://dotat.at/
FORTIES CROMARTY FORTH: WESTERLY OR SOUTHWESTERLY, BECOMING VARIABLE IN
NORTH
FOR A TIME, 3 OR 4. SHOWERS IN NORTH. GOOD.
