Training SpamAssassin

[Exile In Paradise]

On Sat, 2003-12-13 at 14:05, Neil B. Cohen wrote:
> sa-learn --spam --file spambox

Is spambox stored in mbox format?
If so, you might want to add the --mbox command line flag to let it
know.

> on that folder. I had 250+ messages in the folder, but when it finishes,
> it says something like "Learned from 1 message".
> So I have a couple of questions:
> 1) Am I doing something wrong here? How do I get it to process all 250
> messages (or did it and it just didn't tell me?)

That should be fine.
My spam archive (for training spam-assassin) is now over 13,000 posts,
and occasionally I will re-train SA using this archive.

Usually, it will only learn from the new messages I have moved to the
archive after the last time I trained it. It just takes a little time.

> 2) Can I expect to get SpamAssassin to get up to the point of getting
> rid of 90% or 95% of the junk that crosses the Net these days?

It won't take as long as you think to pump the Bayesian hits up to the
95-99% range for new spams.

And, SA quickly gets to the 99.999 or better spam trapping capability.
Its been a long while since I can last remember seeing a spam make it
through to the inbox.

SA is "da bomb!", especially when linked right into Evolution as a
filter.

> For the record, I'm using Evolution to move messages from my input
> folder into the spam folder, and I ran sa-learn under my own account,
> not as root...

I have 2 spam folders.. spam and spam-archive.
I have lots of rules in Evolution to file mailing lists and such...
anything that does not have its own rule usually lands in the Inbox.

So, the last rule in my Evolution filters checks SA like this:
If
   Pipe Message to Shell Command "/usr/bin/spamc -c" does not return 0
Then
  Move to folder "spam" in "local folders"

This way, my Inbox stays clean and junk accumulates in the "spam"
folder... then I can trawl through to make sure nothing I want to save
got accidentally filed as spam, then move all the "processed" spams to
the spam-archive folder.

Then, to train SA, I run this every so often:
sa-learn --spam --mbox --showdots --file \
~/evolution/local/SPAM-Archive/mbox

Hope all of this helps someone.
-- 
Exile In Paradise
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
Url : http://lists.fedoraproject.org/pipermail/users/attachments/20031213/3899b845/attachment-0002.bin