spam levels [was: Including html-tag contents ...]
David Relson
relson at osagesoftware.com
Tue May 13 21:05:08 CEST 2003
At 02:58 PM 5/13/03, Tony L. Svanstrom wrote:
>On Tue, 13 May 2003 the voices made Tony L. Svanstrom write:
>
>TLS> 100 words/tokens, 50 MIN_DEV-eliminated, 30 spammish and 20 hammish
>tokens
>TLS> from the text used; resulting in this many spammish/hammish tokens to
>be used:
>TLS>
>TLS> Spammish Hammish
>TLS> -U1 30 40
>TLS> -U2 30 30
>TLS> -U3 30 20
>TLS> -U4 40 20
>TLS> -U5 50 20
>
> Please ignore the pseudo-math resulting in: 20% == 10 tokens,
>and 30% == 20tokens.
If I understand, the idea is to add a fixed number of tokens (0, 10, or 20)
with scores of 0.0 or 1.0. The number of tokens is _not_ a
percentage. Have I got the idea?
More information about the Bogofilter
mailing list