In data Sat, 26 Jul 2003 22:37:52 -0400
Daniel Carrera <[EMAIL PROTECTED]> scriveva:

> 1)  Use the Bayesian feature of SA so that it can "learn" to
> distinguish spam better.

That would mean to use 

use_bayes       1
auto_learn      1

right?


> 2)  Instead of blindingly strengtening all rules, try to figure out
> some new rules that are likely to catch the spam you are getting and
> not legitimate mail.
> 
> It is best to have many rules with small scores than a few rules with 
> large scores.

How/where can I make my own rules and/or change the existing ones?

Thank you!

-- 
Bonny - Registered Linux User #251752
       --- VB LUG Moderator ---
Brain over - Insert coin.

Attachment: pgp00000.pgp
Description: PGP signature

Reply via email to