Re: [Evolution] EVOLUTION and SPAMASSASSIN

["D. D. Brierton" <darren dzr-web com>]

On Fri, 2003-10-31 at 13:05, Fedele Stabile wrote:

> I have SpamAssassin running as filter on my Evolution,
> but now i want to use sa-learn automatically.
> Any suggestion on how i can operate on Evo?
> AT the moment i have a subfolder SPAM that i examine befor delete ol
> spam mail, but i would automatically use sa-learn whenever i have spam.
> What about creating a filter rule that execute 
> spamassassin -e
> sa-learn
> in sequence?

I don't really see how you can automate the sa-learn process. You have
to specifically tell it whether what it is looking at is spam or
non-spam (ham). Training sa-learn on messages that spamassassin has
already correctly identified as spam is a waste of time.

In order to use sa-learn you first have to train it on a whole bunch of
non-spam. Pick some of you biggest folders (which contain no spam) and
run sa-learn on them like this:

sa-learn --ham --mbox ~/evolution/local/path/to/folder/mbox

Next, in Evolution create a folder called sa-learn. In that folder
create two sub-folders: ham and spam. Now, whenever you download new
mail, move any message that spamassassin incorrectly identifies as spam
into the ham folder, and any spam messages that spamassassin fails to
identify into the spam folder. (Note that this cannot be automated: you
are teaching sa-learn; it's not AI - it doesn't learn all by itself!)
Then periodically run the following two commands (you could even create
cron jobs for them if you wanted):

sa-learn --ham --mbox ~/evolution/local/sa-learn/subfolders/ham/mbox
sa-learn --spam --mbox ~/evolution/local/sa-learn/subfolders/spam/mbox

Afterwards, you can delete the messages in the spam folder and do as you
please with those in the ham folder.

Best, Darren

-- 
=====================================================================
D. D. Brierton            darren dzr-web com          www.dzr-web.com
       Trying is the first step towards failure (Homer Simpson)
=====================================================================