Quick Question

[Chris Wilkes]

On Thu, Mar 24, 2005 at 07:44:56PM -0500, Greg Louis wrote:
> On 20050325 (Fri) at 0021:33 -0000, Jamie Burns wrote:
> > So instead of simply [Non-Spam|Spam] I could have [Family|Friends|Lists|Work]?
> 
> I believe you would have to undertake extensive and difficult
> revisions, and resource consumption as well as uncertainty would
> skyrocket by comparison with the two-class case.

You could try and use different wordlists (ie ~/.bogofilter-family/
~/.bogofilter-friends/ ~/.bogofilter-lists/) and then run your email
through those to see if you got a match.  However that's going to be N
expensive where N == the number of filters you want.

With qmail dot files you could do something like so:
  | condtomaildir ./Maildir/.friends/ /usr/bin/bogofilter -d ./.bogofilter-friends/
  | condtomaildir ./Maildir/.family/ /usr/bin/bogofilter -d ./.bogofilter-family/
etc.  Then maintain the wordlists in each one.

A glaring problem happens if "friends" matches with a 0.95 score but
"family" matches with a 0.99 -- since friends was first in your list the
email is classified as that.  You could get around that by writting
another wrapper to collect all the scores and then put the email in the
correct bucket.

What I've done to classify email is to use dash extensions (for example,
this email is cwilkes-bf at ladro.com) and then filter based on that.

Chris
_______________________________________________
Bogofilter mailing list
Bogofilter at bogofilter.org
http://www.bogofilter.org/mailman/listinfo/bogofilter