help using the new ignore wordlist feature
David Relson
relson at osagesoftware.com
Mon Jun 14 17:38:10 CEST 2004
On Mon, 14 Jun 2004 11:20:22 -0400
Eric Wood wrote:
> ----- Original Message -----
> From: "David Relson"
> > Since the great majority of
> > messages from those lists are ham, my wordlists have a lot of
> > strongly hammish tokens from those lists, i.e.
> >
> > List-Help
> > List-Id
> > List-Post
...[snip]...
>
> Okay, that's making since. Would it be advisable to add?:
>
> head:Date
> head:X-Mailer
> head:User-Agent
Hi Eric,
Use "bogofilter -p wordlist.db" to see if those tokens yield scores that
would matter to you. I'd expect those three tokens to be in most every
message, hence to have neutral scores, hence not be useful in the ignore
list.
Here are my scores for them:
spam good Fisher
head:Date 62636 75249 0.500145
head:X-Mailer 27547 26673 0.553861
head:User-Agent 806 17909 0.051323
The way I determined those tokens should be included was to take a bunch
of "Unsures", build a new wordlist with them, see which ones had high
counts, then select the ones that seemed like they should not be part of
the final result, i.e. select those that I wouldn't include as part of
the ham/spam scoring.
As alwasy, the proper tokens for your lists depend on your mail.
David
More information about the Bogofilter
mailing list