still problem in spam management

[Anne Wilson]

On Friday 08 April 2011 15:21:10 Stéphane Guedon wrote:
> On Friday 08 April 2011 13:34:13 David Relson wrote:
> > On Fri, 8 Apr 2011 08:34:26 +0200
> > 
> > Stéphane Guedon wrote:
> > > On Thursday 07 April 2011 23:16:25 RW wrote:
> > > > On Wed, 6 Apr 2011 09:41:56 +0200
> > > > 
> > > > Stéphane Guedon <stephane at 22decembre.eu> wrote:
> > > > > If in spam but without "___SPAM___" I tell it's really a spam...
> > > > > grep -L "___SPAM___" * | bogofilter -Ns -d /var/lib/bogofilter
> > > > > 
> > > > > 
> > > > > contrary on the ham boxes :
> > > > > grep -l "___SPAM___" * | bogofilter -Sn -d /var/lib/bogofilter
> > > > > 
> > > > > The scripts are not mine, but sounds good !
> > > > 
> > > > How do you prevent the mail being relearned every day? You can
> > > > delete the spam, but what about the ham?
> > > > _______________________________________________
> > > > Bogofilter mailing list
> > > > Bogofilter at bogofilter.org
> > > > http://www.bogofilter.org/mailman/listinfo/bogofilter
> > > 
> > > I will delete old spam when this problem would be solved...
> > > 
> > > I didn't take that into account ! that's good ! but there's so few
> > > false positive I don't think it's really severe.
> > > 
> > > So, ideas ?
> > 
> > I mail incorrectly classified messages to two special addresses on my
> > mail server.  On address is for messages that should have been
> > classified as ham and the other is for messages that should have been
> > classified as spam. On the mail server, a script checks the X-Bogosity
> > 
> > line, then trains bogofilter with the following flags:
> >  -Sn -- for false positives
> >  -Ns -- for false negatives
> >  -s  -- for unsures that are spam
> >  -n  -- for unsures that are ham
> > 
> > Hope this helps,
> > 
> > David
> > 
> > _______________________________________________
> > Bogofilter mailing list
> > Bogofilter at bogofilter.org
> > http://www.bogofilter.org/mailman/listinfo/bogofilter
> 
> So, you "transfer" a copy of the mail ?
> 
> My question is regarding to the fact that I mount my mail server on imap,
> and use sieve, thus, the mail goes directly in the good box according to
> rules (facebook, gentoo, university, spam...). By sending mail to an
> adress, I keep my mail in the good box, delete the spam, and that's it. I
> am right ?
> 
> Should I create a fake adress which I write on porn websites to make it
> send directly spam to me (to the wrong of course...), thus, the training
> should be better ? I think of an adress tartuff at ... to forbid any website
> to understand that it's a spamhole...
> 
> can you show your script ?
> 
> thank you for your help !

I use Bogofilter within Procmail, so although your setup is not identical to 
mine, perhaps my script (created with a lot of help from David) will be useful 
to you.  You'll find it on 
http://userbase.kde.org/KMail/FAQs_Hints_and_Tips#Spam_filtering_on_an_IMAP_server

This does separate out the spam and the unsure mails.  I then have to folders, 
Bogotran_ham and Bogotrain_spam into which I drag the Unsure ones.  Once or 
twice a year :-) I get a definitely wrongly categorised message, and I put a 
copy of that into the training folders.  

I usually wait until I have around 100 spam messages, or until ClamAV tells me 
that one of the messages carries a virus or trojan, then I ssh to my server 
and run

bash /usr/share/bogofilter/contrib/contrib/trainbogo.sh -c -H 
/home/anne/Maildir/.INBOX.Bogotrain_ham/cur/ -S 
/home/anne/Maildir/.INBOX.Bogotrain_spam/cur/

(all on one line, of course).

Bogotrain has served me well over several years.

Anne
-- 
New to KDE Software? - get help from http://userbase.kde.org
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 198 bytes
Desc: This is a digitally signed message part.
URL: <http://www.bogofilter.org/pipermail/bogofilter/attachments/20110408/bf7ff6bf/attachment.sig>