mailscanner at joolee.nl
Fri Nov 16 08:26:06 GMT 2012
Also, Bayes doesn't process images. As you state in your first email, the
rest of the e-mail changes constantly. That way, Bayes can't recognize it.
On 15 November 2012 22:17, Scott Silva <ssilva at sgvwater.com> wrote:
> on 11/15/2012 11:49 AM Max Kipness spake the following:
> >> Get the full spam email with all the headers in a text format pref
> before mailscanner has seen it, then goto pastebin.org and drop it to
> that . The send the link here
> >> We can then copy the email to our systems and see what rules hit on our
> systems to see if any of our setups do a better job
> > Ok, I'll see what I can do. In order to get a copy before it hits
> MailScanner I would have to use MailScanner archiving?
> > Also, what are your thoughts on bayes being at 50% for a lot of spam?
> Would it be best to start over with the bayes database? It would seem that
> by having auto-learning turned on, and a few spam messages get through,
> this could create a domino effect as more and more is learned as non-spam
> when it actually is spam. I continue to send tons of the same type of spam
> through for learning, but I will then see the same type come through with
> 50%. By starting over and feeding 1000k + of legitimate spam, would that
> clean the bayes scoring up?
> > Thanks.
> Autolearn is not very good. I turned it off long ago. Do you do bayes
> maintenance? Like sa-learn --force-expire...
> MailScanner mailing list
> mailscanner at lists.mailscanner.info
> Before posting, read http://wiki.mailscanner.info/posting
> Support MailScanner development - buy the book off the website!
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the MailScanner