Replied-to-sender and CC'ed to the list (I usually reply direct to the
list, not the sender) as Sourceforge has rejected my reply twice.  >:(

Bill Polhemus wrote:
> Since I don't want the stuff appended by SA to be part of the email used to
> train Bayesian, I have to go through each message (I use PINE for this) and
> write out the ORIGINAL message to a separate file which I then use for
> training.

Sounds like you're using "report_safe 1" or "report_safe 2";  where SA
will attach the original spam to a "new" message containing the spam
report.  Personally, I've found it far simpler to use report_safe 0
(which just adds several X-Spam-* headers).  (This also means it's far
easier to "preview" spam in a graphical email client or some webmail
systems;  you don't have to open an attachment.  Which makes it *far*
more usable for ISP customers.  :/  )

With Pine, you could then just save a false negative to a new folder (I
use newspam), then periodically run sa-learn --spam --mbox
{path_to_newspam_folder}.  sa-learn will ignore the headers added by
SA.  Likewise, you can go through your spam folder, save false positives
(if any) to a notspam folder, and then sa-learn that folder.

You *may* have to open the mail folder file in your favourite editor to
remove the "This message is a part of the structure of your mail folder"
message, but I don't think it's too critical as I wouldn't expect to see
such a message actually get sent anywhere.  <g>

I've heard mixed reports on exactly what sa-learn will do with a bare,
out-of-the-box report_safe 1 message;  if for some reason I was forced
to use that mode I'd write a script to unwrap the original message and
drop it in a known location.

-kgd
-- 
<erno> hm. I've lost a machine.. literally _lost_. it responds to
ping, it works completely, I just can't figure out where in my
apartment it is.


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
Spamassassin-talk mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/spamassassin-talk

Reply via email to