#159 managing spam folder
hi, i have read a bunch from the sourceforge/spambayes
pages and if this is a duplicate, then i appologize up front
for not finding it. i think users, me included, need some help
in managing the space consumed by spambayes spam
folder. it would appear from what i have read, that keeping
some spam is helpful for retraining purposes. but from my
experience training is best when the ham/spam training sets
are about balanced -- since i keep ham in lots of folders, it
is very hard to get the right set of folders with a good cross
section of candidates selected for the ham part to balance
the spam folder contents (this could lead to another
discussion of making training easier for the user...later
perhaps). and it is possible for a ham to get lost in the spam
folder (although i quite checking some time ago as it has
never been a productive exercise). so there must be some
balance between all spam since the beginning of time and
an empty spam folder. to begin with i don't know if the
messages should be deleted by age or by spam rating or
some other criteria. one also has to consider what
the "right" amount of spam for training should be (and I
have no idea as an end user). i do think they should go to
the deleted folder and not to the bit bucket . i am thinking
that some high water mark (number of messages to be left
in the spam folder) along with one or more user selecteable
criteria would make an automatic spam folder management
scheme. if you are tempted to write a response which
describes a process i could follow, wouldn't it be better to
codify that response as a new management feature of
spambayes for all to take advantage of? Thanks, Jon
Logged In: YES
user_id=552329
IMO there really isn't any need for any users to keep spam
around (after they have checked it for false positives)
unless they have a research interest in it. If you need to
retrain, it's better to start from scratch and do a
mistake-based training.
See http://entrian.com/sbwiki/TrainingIdeas for more.