Training Spamprobe automatically

Ossi Herrala <[email protected]> Tue, 18 Jul 2006 14:54:31 +0300
Newsgroups gmane.mail.spam.spamprobe.general
Message-ID <[email protected]>
Hello list

I'm a moderator for many mailing lists and we are using Mailman.
Mailman sends e-mail every time a mail is left for moderator and it
containts info about mail left for moderation and the body of that
mail.

I'm using "spamprobe receive" for every incoming mail. So, I'm
training spamprobe with moderation notices. And when spam is left for
moderation, spamprobe learns the message as good.

So, my database is polluted with spammy tokens marked as ham. Which
might degrease accuracy of Spamprobe.


Now I have put spamprobe in score mode to receive mails. But I should
train my database all the time, shouldn't I? But how should I train
it?

I use procmail to filter my mails to different Maildir boxes. The good
mails are in Mail/<box name>/cur and spam is in Mail/spam/cur [1]. I
also check subject of every received spam (still doing it, but I
haven't never found any ham in my spam..) and mark it as read.

I think I could run a script from crontab to "spamprobe auto-train"
all read ham and spam, but some of my Maildir boxes are huge. It would
eat a processing power every time and still check same mails over and
over again.

Anyone else in the same situation? Any scripts or suggestions about
training the database?


[1] New mail arrives to Mail/<box name>/new and is transferred to
    Mail/<box name>/new when read.

-- 
Ossi Herrala, OH8HUB
PGP: 0x78CD0337 / D343 F9C4 C739 DFFF F619  6170 8D28 8189 78CD 0337 

Hi! I am a .signature virus.  Copy me into your .signature to join in!

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys -- and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV

_______________________________________________
Spamprobe-users mailing list
Spamprobe-users-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org
https://lists.sourceforge.net/lists/listinfo/spamprobe-users
signature.asc (application/pgp-signature, 185 B)
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.4 (SunOS)

iD8DBQFEvMv37NFsifUzPh0RAhxvAJ9mQoD3elgf8gw5GkCIqbIjr6YHogCfVgpk
GOMERxvc5DlECtOXmCTLAHU=
=upcT
-----END PGP SIGNATURE-----