Bayes?

[Desai, Jason]

Very interesting!  Two questions:

1)  I'm assuming that your users just forward their emails to these
addresses.  Will it make a difference to the bayes filter that the message
has been forwarded, or is it just looking at the body of the message?

2)  Do you have to worry about some kind of locking of the bayes database
when running sa-learn, so that sa-learn is not updating it while MailScanner
is calling spamassassin to use it at the same time?

Thanks.

Jason

> -----Original Message-----
> From: Julian Field [mailto:mailscanner at ECS.SOTON.AC.UK]
> Sent: Friday, March 21, 2003 9:57 AM
> To: MAILSCANNER at JISCMAIL.AC.UK
> Subject: Re: [MAILSCANNER] Bayes?
>
>
> At 14:51 21/03/2003, you wrote:
> >Just qurious about spamassassins bayes function in
> connection with MS.
> >Is MS utilizing the bayes function automatically
>
> Yes.
>
> >  or do you have to
> >configure something to make it work?
>
> No.
>
> >  If so what, and do you have to
> >run a "learning session" with it?
>
> No.
>
> >  That is, do you have to run some
> >definitly SPAM emails and some non SPAM emails through it?
>
> No.
>
> MailScanner is using the bayes function automatically.
> Uniquely, the bayes
> engine in SpamAssassin is "self-learning"; it uses the other rules to
> identify messages which have either a very high score or a
> very low score,
> and it continuously feeds them to the bayes engine itself without you
> needing to do anything.
>
> If you want to teach it when it gets it wrong, you can have it run the
> "sa-learn" script to learn about particular messages. I have set up 2
> addresses here, "spam" and "notspam". Their mailboxes live on the main
> MailScanner server, and people can just redirect wrongly-classified
> messages to one of the addresses. Then once an hour the
> script below is run
> by cron to teach the bayes engine about the messages it got
> wrong. For the
> script below, I have copied SpamAssassin's "sa-learn" script into the
> MailScanner bin directory.
>
> You should also run a nightly cron job that does a "sa-learn
> --rebuild" as
> well, to do all the housekeeping the Bayes engine requires.
>
> #!/bin/sh
>
> SPAM=/var/mail/spam
> NOTSPAM=/var/mail/notspam
>
> LOGFILE=/var/log/learn.spam.log
> PREFS=/opt/MailScanner/etc/spam.assassin.prefs.conf
> SALEARN=/opt/MailScanner/bin/sa-learn
>
> date >> $LOGFILE
> if [ -f $SPAM ]; then
>    BOX=${SPAM}.processing
>    mv $SPAM $BOX
>    sleep 5 # Wait for writing current message to complete
>    $SALEARN --prefs-file=$PREFS --spam --mbox $BOX >> $LOGFILE 2>&1
>    rm -f $BOX
> fi
>
> if [ -f $NOTSPAM ]; then
>    BOX=${NOTSPAM}.processing
>    mv $NOTSPAM $BOX
>    sleep 5 # Wait for writing current message to complete
>    $SALEARN --prefs-file=$PREFS --ham --mbox $BOX >> $LOGFILE 2>&1
>    rm -f $BOX
> fi
>
> --
> Julian Field
> www.MailScanner.info
> MailScanner thanks transtec Computers for their support
>