Bayesian Learning and forward to as spam action

[Julian Field]

At 23:12 24/11/2003, you wrote:
>Hello Julian,
>
>Monday, November 24, 2003, 3:40:12 AM, tu as écrit:
>
>JF> At 02:21 24/11/2003, you wrote:
> >>Hello MS users,
> >>Ms on a smtp gateway with just 2 local users (spam and nospam)
> >>all mailboxes are  in exchange
> >>
> >>I want tu use a Bayesian Learning Script.
>
>JF> SpamAssassin will already auto-learn from messages it is convinced are
>JF> either spam or not spam. You don't have to forward messages using the 
>"spam
>JF> actions" setting, it is already happening whenever it is appropriate 
>to do so.
>
>Thanks for the response,
>my question is :
>How can we train MS about false positive and negative ? I ubderstand
>now that I don't have to worry about the learning process of spam/not
>spam but howe about False + and false -. I don't understand how he can
>learn the goob way when he was wrong :-)

I have a script which I have posted to this list several times before. It's 
called something like learn.spam and works from the contents of a couple of 
mailboxes, "spam" and "notspam".

It's in the list archives.
http://www.jiscmail.ac.uk/cgi-bin/wa.exe?A2=ind0304&L=mailscanner&P=R12099&I=-1&X=749CC23485FE24E21C&Y=jkf%40ecs%2Esoton%2Eac%2Euk 

-- 
Julian Field
www.MailScanner.info
MailScanner thanks transtec Computers for their support

PGP footprint: EE81 D763 3DB0 0BFD E1DC  7222 11F6 5947 1415 B654