how to use bogofilter efficiently

how to use bogofilter efficiently

[Uwe Brauer]

Hello 

I changed by .gnus spam setting from 
(setq spam-use-BBDB-exclusive  t)
 (setq spam-use-BBDB t)

To 

(setq  spam-use-bogofilter t)
this resulted that no message was filtered so I set

(setq spam-bogofilter-database-directory "/home/oub/.bogofilter")
and generated a database with
bogofilter -s SPAM
bogofilter -n INBOX

However there a still false negatives, so the only possibility I see
is to move them to the folder SPAM and run
bogofilter -s SPAM
but isn't there a better more efficient way to do it.

(I have disabled fetchmail, spamassin and my procmail filter)

Thanks


Uwe Brauer 

Re: how to use bogofilter efficiently

[Berthold Höllmann]

Ted Zlatanov <tzz@lifelogs.com> writes:

> On Wed, 22 Oct 2003, oub@mat.ucm.es wrote:
>
>> However there a still false negatives, so the only possibility I see
>> is to move them to the folder SPAM and run
>> bogofilter -s SPAM
>> but isn't there a better more efficient way to do it.
>
> You can add a bogofilter spam exit processor to the "SPAM" group.  It
> will run the spam articles through bogofilter as you are describing
> it.  It's not perfect yet, but should work for you.

And how can a register my Ham as Ham for bogofilter? I'd like all new
mail not marked as Spam to be registered as Ham.

Regards
Berthold
-- 
Dipl.-Ing. Berthold Höllmann   __   Address:
hoel@GL-Group.com           G /  \ L Germanischer Lloyd
phone: +49-40-36149-7374    -+----+- Vorsetzen 32/35    P.O.Box 111606
fax  : +49-40-36149-7320      \__/   D-20459 Hamburg    D-20416 Hamburg

Re: how to use bogofilter efficiently

[Ted Zlatanov]

On Thu, 23 Oct 2003, hoel@gl-group.com wrote:

> Ted Zlatanov <tzz@lifelogs.com> writes:
> 
>> On Wed, 22 Oct 2003, oub@mat.ucm.es wrote:
>>
>>> However there a still false negatives, so the only possibility I
>>> see is to move them to the folder SPAM and run bogofilter -s SPAM
>>> but isn't there a better more efficient way to do it.
>>
>> You can add a bogofilter spam exit processor to the "SPAM" group.
>> It will run the spam articles through bogofilter as you are
>> describing it.  It's not perfect yet, but should work for you.
> 
> And how can a register my Ham as Ham for bogofilter? I'd like all
> new mail not marked as Spam to be registered as Ham.

Use the bogofilter ham exit processor.  The question then becomes,
what is and isn't ham?  You have to not only add the bogofilter ham
exit processor, you should also customize the ham-marks group
parameter if the default is not right for you.  For me, only ticked
(!) articles are ham (so ham processing doesn't happen for all
articles in a ham group), but you may want to expand that if you want
more non-spam mail to be ham.  There are also some settings on whether
ham should be processed in all groups and spam groups:
spam-process-ham-in-nonham-groups and spam-process-ham-in-spam-groups;
you can move ham around with ham-process-destination and spam with
spam-process-destination...

All this is in the Gnus manual, let me know if any part of it is
unclear.

Thanks
Ted

Re: how to use bogofilter efficiently

[Ted Zlatanov]

On Thu, 23 Oct 2003, oub@mat.ucm.es wrote:

On 22 Oct 2003, tzz@lifelogs.com wrote: 
>> You can add a bogofilter spam exit processor to the "SPAM" group.
>> It will run the spam articles through bogofilter as you are
>> describing it.  It's not perfect yet, but should work for you.

> I am not sure I understand: I split my incoming mail on the imap
> server with the bogofilter into MAILBOX and the SPAM group.  This
> splitting is based on my database.  Now, since this splitting is far
> from optimal, I want to actualize my database. Should I move the
> relevant from my MAILBOX to the SPAM box?  Otherwise how does
> bogofilter know about the new database?

Because your spam-splitting is done on the server, you can't really
train bogofilter on a client.  The best you can do is move articles
to a "train" group and then run bogofilter on it on the server
every now and then.  spam.el has some facilities for moving articles
around.

You could hack the source to make a remote-training-bogofilter
processor, but that's probably not something you're interested in
doing, and definitely isn't something I have the time to do :(

> In addition: could I combine (setq spam-use-BBDB-exclusive t) and
> (setq spam-use-bogofilter t)?  Like first the mail is split into the
> MAILBOX (only bbdb members in) and SPAM, spam and unclassified, (*)
> then a bogofiler runs on this splitting mail into REALSPAM and
> HARMLESS or so.

Not in that order.  You can do a specific order thus in the fancy
split rules:

--> set spam-split-group to "REALSPAM"
(: spam-split 'spam-use-bogofilter)
--> set spam-split-group to "MAYBESPAM"
(: spam-split 'spam-use-BBDB-exclusive)
("mail")

If you do it this way, you'll have to set spam-split-group with a
function outside of spam-split.  Let me know if this makes some
sense.

Ted