Problems with BAYES

[Tieum <ti...@free.fr>]

   Hello,

I use spamassain behind amavisd-new. I get two prblems with BAYES:

1) my imap folders are scanned everyday some for HAM some for SPAM, everyday 
ssa-learn tells it learned from all messages although some are not new

2) some of my messages are checked for BAYES (and therefore have BAYES_nn 
token) and some other are not. WHy aren't they all checked ?


  Thanks,


  Matthieu

Re: Problems with BAYES

[Tieum <ti...@free.fr>]

Le Lundi 23 Mai 2005 06:09, vous avez écrit :

> >1) my imap folders are scanned everyday some for HAM some for SPAM, 
everyday ssa-learn tells it learned from all messages although some are not 
new

> That's weird. Is your bayes_seen file getting deleted?

It does not seem so


> >2) some of my messages are checked for BAYES (and therefore have BAYES_nn 
token) and some other are not. Why aren't they all checked ?

> What version of SA do you use? 2.6x? If so, that's normal. A perfect 50/50  
score from bayes will just cause no bayes rules to hit. 

I use version 3.0.3-3.fc3

How can I solve this problems or at least diagnose where they come from ?


  Thanks,

Matthieu

Re: Problems with BAYES

[Matt Kettler <mk...@comcast.net>]

At 10:20 AM 5/21/2005, Tieum wrote:
>I use spamassain behind amavisd-new. I get two prblems with BAYES:
>
>1) my imap folders are scanned everyday some for HAM some for SPAM, everyda=
>y=20
>ssa-learn tells it learned from all messages although some are not new

That's weird. Is your bayes_seen file getting deleted?


>2) some of my messages are checked for BAYES (and therefore have BAYES_nn=20
>token) and some other are not. WHy aren't they all checked ?

What version of SA do you use? 2.6x? If so, that's normal. A perfect 50/50 
score from bayes will just cause no bayes rules to hit.