[Spambayes] RE: Eliminating many malformed spams

[Steve Holden]

[Barry]
> On Thu, 2003-06-19 at 12:30, Steve Holden wrote:
> > Hi. Time for my latest observation, which is that while spambayes is
> > great (especially when integrated with Outlook, for this
> Windows user)
> > there seem to be a number of automated mailing systems that generate
> > malformed messages. I am beginning to suspect they may do it
> > deliberately to foil attempts at despamming.
>
> If that was truly the case, and we could count on proper RFC
> compliance
> by all legitimate mailing systems, then you could just throw all
> non-parseable message into the spam folder and be done with it, right?
>
Yes, and I've even considered doing that, but I'm reluctant to trust
that all legitimate mailers will be RFC-compliant. I'm also reluctant to
draw conclusions from any particular characteristics of the
malformation - once you start doing that, spammers will start using it
as a way to misinform your filtering process.

> Still, I understand there are times when you really want to be able to
> parse any random stream of characters <wink> into a mail message
> object.  Anthony Baxter is hard at work on a more lenient parser that
> should improve the rate of parseable messages.

That's good news. More strength to his arm! At present I'm getting about
300 mails per day. About twenty of these are legitimate, I usually get
about ten "unsure" (of which 98% are spam) and twenty or so unparseable.
Sn I'm hoping in the meantime that my little tweak will reduce my
workload somewhat.

>                                             I think any
> application
> though has to be prepared to handle un-parseable messages, as we'll
> never get it to 100%.
>
Correct.

regards
--
Steve Holden                                 http://www.holdenweb.com/
Python Web Programming                http://pydish.holdenweb.com/pwp/