[Spambayes] Obvious Spam Missed...
Mark Hammond
mhammond at skippinet.com.au
Tue Sep 16 19:30:53 EDT 2003
> From: Meyer, Tony [mailto:T.A.Meyer at massey.ac.nz]
> > Since it's so easy to suck existing mailboxes into the
> > Outlook plugin, perhaps it should warn the user if the
> > ham:spam ratio gets too out-of-whack.
>
> I think this is probably a good idea. What "too
Outlook2000\dialogs\dialog_map.py has the following class:
class WizardTrainingStatusProcessor(ControlProcessor):
def Init(self):
bayes = self.window.manager.classifier_data.bayes
nspam = bayes.nspam
nham = bayes.nham
if nspam > 10 and nham > 10:
msg = "SpamBayes has been successfully trained and configured.
" \
"You should find the system is immediately effective at "
\
"filtering spam."
else:
msg = "SpamBayes has been successfully trained and configured.
" \
"However, as the number of messages trained is quite
small, " \
"SpamBayes may take some time to become truly effective."
win32gui.SendMessage(self.GetControl(), win32con.WM_SETTEXT,
0, msg)
This would be a simple place to add this status information (it shows up in
the "manager" dialog). Seeing as I picked the number "10" from my anus, I
see no reason we couldn't do the same with magic cutoffs for an inbalance
warning <wink>
Mark.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: winmail.dat
Type: application/ms-tnef
Size: 2344 bytes
Desc: not available
Url : http://mail.python.org/pipermail/spambayes/attachments/20030917/d655cf2e/winmail.bin
More information about the Spambayes
mailing list