First: somebody please let me know if magnets ONLY
sort matching messages into the corresponding bucket
(as I assume), OR if they add the message's text to the
classification corpus.
I'd like to be able to:
+ Create a spam-attractor email address (or three), and
put them in places where they're likely to get picked up
by spammers.
+ Create magnet(s) for those email addresses that mark
them as spam AND automatically run them through the
classification system, as if I'd manually classified them
as spam.
Logged In: YES
user_id=848514
great idea!
Logged In: YES
user_id=578491
This is a bad idea because it means you'll be adding tons of
information to the spam corpus. That means that POPFile is
going to be biased towards thinking all your mail is spam,
which it isn't really.
Just use the train on errors methodology.
John.
Logged In: YES
user_id=697183
I'll modify my request:
Make the proposed "reclassify" magnet only add any
mis-classified messages to the corpus.
Reasoning: Everything sent to my spam-catcher address is
spam. If PopFile disagrees, I WOULD manually re-classify
the error, but I'm lazy and would rather just set up a rule
to do that for me.
Logged In: YES
user_id=900207
I am unsure, but believe that the magnets do not add to
corpus, but it would be nice to have an option for magnets to
automatically classify, check the classification, and then
reclassify if wrong.