Bogofilter seems to not be working

[Michael D Richards]

David Relson wrote:

> As you say, this process is learning and extending bogofilter's 
> vocabulary.  It also assumes bogofilter is doing a good job (which it 
> can do).  However, bogofilter can not be all knowing, so there will be 
> messages that are incorrectly classified.  When bogofilter is first 
> being used, the wordlists are small, and bogofilter's accuracy is at 
> its worst.  When using "-u", the sysadmin _must_ monitor what 
> bogofilter is doing.  When bogofilter makes a mistakes, the sysadmin 
> needs to notifiy bogofilter and have the message removed from one 
> wordlist and added to the other.


Just to chime in with a little personal experience...

I have setup a system where -u is used starting with zero corpora. 
Bogofilter assumes everything is "good" at first and as users start 
fixing the false negatives, accuracy builds *very* rapidly. Under this 
system I have only seen one false positive, and chances are that one 
would have been classified that way even under a highly trained and 
mature instance.

In other words, even when bogofilter's accuracy is "at its worst", it 
works very well and -u can be very useful if the surrounding system is 
well thought out and you make it very easy for users to fix errors on 
their own.

Michael~
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.bogofilter.org/pipermail/bogofilter/attachments/20030325/8ca27f4e/attachment.html>