Re: UNS: Re: crm filter cleanup on productive system
Ger Hobbelt <[email protected]>
| Newsgroups | gmane.mail.spam.crm114 |
|---|---|
| Message-ID | <[email protected]> |
On Sun, Feb 22, 2009 at 10:02 PM, Ger Hobbelt <[email protected]> wrote: > For completeness, here's how crm114 trains 'out of' instead of 'into': > you can specify the training option <refute> (see QUICKREF.txt) with a > classifier 'train' command: that's the signal crm114 needs to know you > want it to train in a way which is saying 'this does /not/ belong in > here', while regular training is a way to teach the classifier that > 'this does belong in this category'. For all of those who say: hey! vanilla mailfilter.cf and mailtrainer.crm already /have/ <refute>, so what's he about? Well, mailtrainer / mailfilter.cf has this (snippet): # Set to empty // if you do not want mailtrainer to also 'refute'-train messages # when initial (regular) training didn't deliver the desired result (yet). # # Default when not specified: SET. # #:do_refute_training: /SET/ :do_refute_training: // but that ONLY means, when SET, mailtrainer will <refute> train after a 'regular' train operation when that regular train instruction didn't produce the desirable result. That's not double-sided training as winnow needs it: that classifier performs best when you /always/ train both regular and <refute>. The significant difference being 'always' versus 'only when' (which, by the way, is 'rather seldom' for the usual test sets). -- Met vriendelijke groeten / Best regards, Ger Hobbelt -------------------------------------------------- web: http://www.hobbelt.com/ http://www.hebbut.net/ mail: [email protected] mobile: +31-6-11 120 978 -------------------------------------------------- ------------------------------------------------------------------------------ Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco, CA -OSBC tackles the biggest issue in open source: Open Sourcing the Enterprise -Strategies to boost innovation and cut costs with open source participation -Receive a $600 discount off the registration fee with the source code: SFAD http://p.sf.net/sfu/XcvMzF8H