[Spambayes] Randomized Spam Beating SpamBayes

[skip at pobox.com]

    Shawn> Thanks Skip. While I can find nowhere that has a Tokenizer
    Shawn> section, the "Outlook.ini" file in the AppData folder looks like
    Shawn> it might do the trick. 

I'd be suspicious of that.  Do a find on your disk for
"default_bayes_customize.ini".  Failing that, try searching for files with
"bayes" in the name.  The default distributed version starts with this
comment:

  # This is the INI file for the *Bayes Engine* as used by the Outlook addin.  
  # It is NOT where configuration information is stored for the addin - see 
  # "[Profile Name].ini" for these settings.

Further down it has this comment about which copy it finds takes precedence:

  # This file exists in the SpamBayes program directory, and may optionally
  # exist in the SpamBayes data directory.  If you wish to make changes to
  # the default options, setting the option in a file in your data directory
  # will persist even when SpamBayes is upgraded.  If you change the version in
  # the SpamBayes program directory, it will be upgraded along with SpamBayes, 
  # so your changes will be lost.  Options in the data directory file have 
  # precedence over the app directory file.

Once you find it, just add the options I mentioned to the [Tokenizer]
section and restart.

    Shawn> I've added the settings below to that file. I'll let you know in
    Shawn> a few days if it made any difference.

Were you the person with, like, 60,000 spams and a similar number of hams in
your training set?  Maybe try retraining from scratch.  I have a total of
about 400 emails in my training set and it works fine.

Skip