[Spambayes] How does SpamBayes training count "spam" and "good" mail

Tony Meyer tameyer at ihug.co.nz
Wed Nov 10 22:34:43 CET 2004

> Anyone know how SpamBayes training counts "spam" and
> "good" mail? In the SpamBayes 1.0 for Outlook, I designated
> initial training to look at about 300 "spam" messages
> in a SPAM folder and about 300 "good" messages in a GOOD
> folder, checkmarked it to rebuild entire database, and it
> comes back with "Completed training with 91 spam and 9 good
> messages" and the Training Database Status tells me the
> "Database only has 9 good and 91 spam - you should consider
> performing additional training." Can anyone tell me why it
> doesn't see the other 500 messages I gave it? What am I missing
> here?

The log file <0.5 wink>.  Seriously, that'll explain why messages were not
trained.  Wild guesses:

  1.  Some of the mail has never been received (drafts, sent items, ...)
  2.  Some of the mail are not what Outlook calls IPM.Note (regular
messages), for example some Exchange autogenerated messages.
  3.  There was a periodic problem accessing the folder (i.e. run scanpst
over it if it's a PST, or check the connectivity if it's

If these don't apply, please feel free to send in a copy of the log file for
the period in which you tried this, and we'll try and explain whatever
errors are there.


Please always include the list (spambayes at python.org) in your replies
(reply-all), and please don't send me personal mail about SpamBayes.
http://www.massey.ac.nz/~tameyer/writing/reply_all.html explains this.

More information about the Spambayes mailing list