ACME Labs mail filtering tutorial

[David Relson]

On Wed, 25 May 2005 18:18:40 -0700
Jef Poskanzer wrote:

> I've been working on this for a few months, and just published it today:
> 
>     http://www.acme.com/mail_filtering/
> 
> Bogofilter is mentioned in the Bayesian section.
> ---
> Jef
> 
>        Jef Poskanzer  jef at mail.acme.com  http://www.acme.com/jef/

Hi Jef,

Nice work.  I've not read everything, but what I have read looks very
nice.  It makes my graphing efforts look amateurish :->

Do remember that bogofilter's default values are deliberately on the
conservative side to cut the number of false positives.  The natural
side effect of this is to increase the number of false negatives.  The
option is always present to change the default values of the spam and
ham cutoffs.  This can be done by choosing values that "seem right", by
choosing cutoffs based on scores (for false negatives and positives),
and by using bogotune's grid search for the best values it can find.

Regards,

David