Moderator messages seen as spam

[Anne Wilson]

Reply to All / Reply to List
On Friday 23 April 2010 17:34:45 Jonathan Kamens wrote:
> If you train and tune bogofilter properly, that should prevent it from
> classifying the moderator messages as spam.
> 
> The problem is that this will also make bogofilter slightly worse at
> recognizing real spam, since it will pollute your ham word list with the
> spammy words in the moderator messages.
> 
Precisely.
> 
> There are two ways you could potentially improve the situation:
> 
> 1.       Are you using tri-state filtering, i.e., "ham, spam or unsure"
> rather than just "ham or spam"?

Yes.  The mail is filtered by bogofilter rules within the procmail recipes.  
When I get a 100 or so new spam I train bogofilter on them, using 

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/

The results are good - I get very few 'problems' - but then I've been using 
bogofilter with the same corpus for some years.

> 2.       I've got my bogofilter configuration set up so that email messages
> with certain Subject lines bypass bogofilter completely.  I bypass
> bogofilter for messages that I know are ham but which tend to contain
> spammy words in them.  If it is possible for you to set this up, then you
> can configure bogofilter to ignore moderator messages completely, thus
> neither incorrectly sending them to your spam folder nor polluting your
> ham word list with spammy words.
> 
I don't see how I can by-pass bogofilter and still have procmail deliver the 
mail to the correct folder - but I'll give it some though.  I'd like to filter 
on a header content - "X-List-Administrivia: yes" which should solve the 
problem, but to work it would have to be in the bogofilter script.  I 
documented my script on 
http://userbase.kde.org/KMail/FAQs_Hints_and_Tips#Spam_filtering_on_an_IMAP_server

If you have ideas on how I could add that filter, I'd love to hear them.  I 
guess that it need something like 'if header contains X-List-Administrivia: 
yes then X-bogosity=ham'  - Could that be done?

Anne
> 
> 
> Jik
> 
> _______________________________________________
> Bogofilter mailing list
> Bogofilter at bogofilter.org
> http://www.bogofilter.org/mailman/listinfo/bogofilter
-------------- 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/20100423/d9513c62/attachment.sig>