[Spambayes] Many users on domain coming up as "possibly spam"

[Kenny Pitt]

This has already been tried. It was called the
"experimental_ham_spam_imbalance_adjustment" option, and it did exactly what
you suggest. It computed the imbalance ratio and used that ratio to adjust
the weights of the scores.
 
Unfortunately, it proved to be such a miserable failure that it was removed
from the source. It initially seemed like a logical approach, but there is
apparently something inherent in the mathematics that defies normal logic.
<wink>
 
-- 
Kenny Pitt
 

  _____  

From: Coe, Bob [mailto:rcoe at CambridgeMA.GOV] 
Sent: Wednesday, October 20, 2004 8:34 AM
To: Kenny Pitt
Cc: spambayes at Python.org
Subject: RE: [Spambayes] Many users on domain coming up as "possibly spam"


I don't understand why this very simple and well understood problem is so
resistant to solution. Since Spambayes is obviously aware of the imbalance
(or it couldn't have quoted the numbers in the log), why can't it simply
discount each spam token by a factor of 64? (Make it optional so that those
of us who don't experience the problem don't have to get involved.)
 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.python.org/pipermail/spambayes/attachments/20041020/7d260386/attachment.htm