With any luck, you're going to 'educate' the filter very seldomly (if you properly trained it at the beginning). See the previous part of the series for details.
The only standalone e-mail client with a fully integrated bayesian filter is Thunderbird, as far as I know. But its native filter isn't very good, in my opinion (1-word tokens only, I think).
KMail doens't offer the possibility of linking a user-supplied script/filter to an icon in the toolbars, otherwise it would have been a snap.
All of the building blocks are in place, anyway, and as soon as KMail implements that possibility (probably with KDE 3.2, going beta shortly) putting it all together will be a no-brainer.
For the time being, may be Sylpheed is a better choice if you need thunderbird-like operations.
Learn about Linux
Download Linux
Get Linux help
Re:How do we "educate" the filter?
Posted by: Corrado on October 31, 2003 01:36 AMThe only standalone e-mail client with a fully integrated bayesian filter is Thunderbird, as far as I know. But its native filter isn't very good, in my opinion (1-word tokens only, I think).
KMail doens't offer the possibility of linking a user-supplied script/filter to an icon in the toolbars, otherwise it would have been a snap.
All of the building blocks are in place, anyway, and as soon as KMail implements that possibility (probably with KDE 3.2, going beta shortly) putting it all together will be a no-brainer.
For the time being, may be Sylpheed is a better choice if you need thunderbird-like operations.
#