SA, Bayes, and You (erm, Me)

[Eric F Crist]

On Jan 11, 2005, at 11:03 PM, dave wrote:

> Hi,
>     Can you tell me how you got to get bayes updated and trained and 
> those
> headers? I'm using MailScanner which invokes sa and so forth and i am 
> not
> seeing the detail nor am  getting the results you are.
> Thanks.
> Dave.

You can train the Bayes filter by using the sa-learn program.  See man 
1 sa-learn for more specific about how to use it.  Basically, you take 
a minimum of 200 messages known to be spam (I use mbox format) and run 
it through such as:

# sa-learn --showdots --mbox --spam /path/to/file

You also need a minimum of 200 ham (non-spam) messages, using the 
following command (again, mbox format):

# sa-learn --showdots --mbox --ham /path/to/ham/file

SpamAssassin should learn on it's own and will begin invoking the Bayes 
filters once it's had the minimum of 200 messages of each spam/ham.  If 
you want more accurate results, feed it more mail.  I save spam I get 
and try to run a bunch of it through every few months, and I do the 
same with ham.

HTH
_______________________________________________________
Eric F Crist                  "I am so smart, S.M.R.T!"
Secure Computing Networks              -Homer J Simpson
-------------- next part --------------
A non-text attachment was scrubbed...
Name: PGP.sig
Type: application/pgp-signature
Size: 194 bytes
Desc: This is a digitally signed message part
Url : http://lists.freebsd.org/pipermail/freebsd-questions/attachments/20050112/79cd180c/PGP.bin