More than one goodlist/spamlist
David Relson
relson at osagesoftware.com
Wed May 7 23:28:00 CEST 2003
At 05:19 PM 5/7/03, Chris Wilkes wrote:
>On Wed, May 07, 2003 at 11:19:49AM -0700, Jef Poskanzer wrote:
> > Is there a way to tell bogofilter to score based on auxiliary .db files
> > as well as the regular pair? I don't see this option.
> >
> > I'm thinking it would be good for a multi-user setup - have a centralized
> > spamlist continuously trained on the mail to some spam-trap users, or
> > by a feed from spamarchive.org or something. Each user would use their
> > own .db files plus the centralized one. Any registrations they do would
> > only go to their own files, of course.
>
>Or how about expanding the idea of Bogofilter to beyond spam so that it
>could classify any mail sent in.
>
>If you had a generic email address like problems at example.com that
>received all incoming trouble ticket emails, and you had setup your
>directory structure like so:
> ./Maildir/
> downedpowerline/
> catuptree/
> gaslinebreak/
> needtoclassify/
>within each of those directories there is a "good" database list. Then
>compare the email's spam score against each one of those directories and
>then filter it to the proper one.
>
>So after a couple of emails bogofilter will be able to sort out if the
>person's cat went up the tree or if they're reporting a downed power
>line, as those keywords ("I smelled some rotten eggs at 14th and
>Vine" vs "there's sparks all over the road!!!") will be added to the
>approriate list.
>
>That's pretty extreme, but it could be useful in a bug reporting tool,
>where you would get a lot of reports of UI errors and you want to sort
>them out.
>
>Chris
Chris,
Bogofilter could easily do that. It doesn't even need to be modified if
you don't mind building multiple wordlists and running it multiple times.
All it takes is a training corpus for each decision and an appropriate
script (or recipe) for running bogofilter with each trouble ticket.
David
More information about the Bogofilter
mailing list