On Wednesday 16 June 2004 23:02, Bruno Broedner wrote:
> Hi!
>
> I receive nearly all spam in English, but also some in German. For the
> English spams the Bayes works fine, because it has autolearned a lot in
> English. But for German spams the Bayes-scores pull down the results and
> spam gets through.
>
> Is there anything I can do against that phenomen?
>
> This is the result for a porn-spam in German:
>
>      X-Spam-Status: No, hits=2.7 required=3.0 tests=BAYES_10,
>      RAZOR2_CF_RANGE_51_100,RAZOR2_CHECK,TDE_FM_BP_KONTAKTE autolearn=no
>      version=2.63
>
> Razor and one German rule (TDE_FM_BP_KONTAKTE ) give their scores, but
> BAYES_10 pulls it back down.
>
> Thanx
> Bruno

Feed the german spam to sa.learn and after a few such feeds you will
find it scores it properly.

-- 
_____________________________________
John Andersen

Attachment: pgp9F9qc6pLZe.pgp
Description: signature

Reply via email to