RE: [AMaViS-user] How to use sa-learn to manually train?

[SamW <sw...@sv...>]

Thanks for pointing out, Yes it was a typo, and should be corrected to, 

/usr/local/bin/sa-learn --ham --mbox /var/mail/notspam

Our issue in training is that all our users are accessing their mail
using a POP-client. So there is no way for them to create folders and
submit them. What if they just forward their spam and non-spam messages
so that SA can do a  sa-learn from cron? This should work, but my
concern is 

-	Will SA read the information from the message header such as the
forwarder's email address and after the learn process BAYES filter will
consider the forwarder as a SPAMMER? 
-	Is anyone outthere using a setup to feed the spam to BAYES? when
users have only pop3 access to email?

Of cource my addresses sp...@my... and no...@my... are
both listed under spam_lovers in /etc/amavisd.conf.

- SamW

> -----Original Message-----
> From: Ricardo Stella [mailto:st...@ri...]
> Sent: Monday, April 12, 2004 9:20 AM
> To: SamW
> Cc: ama...@li...
> Subject: Re: [AMaViS-user] How to use sa-learn to manually train?
> 
> 
> 
> SamW wrote:
> >
> > Here is how I use the sa-learn to train the bayes filter in version
SA
> > 2.63. I am using amavisd-new-p9 / sa 2.63 / sendmail
> >
> > I asked our users to send a mail to sp...@my... if the users
see a
> > spam message. If they see a non spam (ham) message they send a mail
to
> > no...@my.... Note that the messages that the user receives
is
> > already tagged with X-Spam-* headers
> >
> > I then run following 2 commands, to train the sa 4 times a day
through a
> > crontab.
> >
> >         /usr/local/bin/sa-learn --spam --mbox /var/mail/spam
> >         /usr/local/bin/sa-learn --spam --mbox /var/mail/notspam
>                                    ^^^^^^                ^^^^^^^^^
> 
> Shouldn't that be 'ham' instead ?  A typo, but a crucial one...
> 
> Second, your users should only use the bounce or redirect messages
> option.  Do not have them forward their mail to an account, as some
> clients do it differently.  Some forward them as attachments, others
> inline, others quoted...
> 
> Your other option is for them to file or save the messages in a spam
or
> ham folder.  A script could visit all of these folders and train sa
> accordingly.
> 
> My .02...
> 
> My .02...