[Spambayes] Question about the Outlook add-in and theclassification of already filtered messages.

[Kenny Pitt]

To answer your summary question, running "Filter messages" with the "Score
messages, but don't perform filter action" will properly rescore all the
messages in the selected folders.  You can also use the "Show spam clues"
item to check the score of the currently selected message.
 
Now, why are the scores dropping instead of increasing?  You say "I get 10
new spammy messages in the junk folder, and I trained the 3 lowest scoring
spam as spam".  Did you really mean that you got 10 messages in your Unsure
folder?  If you really did mean that they are in the junk folder then the
only training option available is "Recover from Spam" which trains that the
message is good. If you trained the messages with "Recover from Spam" then
the scores of any similar messages would definately drop.
 
If you are in the Unsure folder and using "Delete as Spam" to train, then it
is not obvious why other spam scores would decrease.  If this happens
consistently then the best thing to do would be to capture the "Show spam
clues" output of one of these messages before training a similar one, and
then again afterwards.  You can then send both sets of clues to this list
and it might give us a better idea of what is going on.  Also, sending your
most recent logfiles as described in the Troubleshooting Guide could also be
helpful.
 
-- 
Kenny Pitt
 

  _____  

From: spambayes-bounces at python.org [mailto:spambayes-bounces at python.org] On
Behalf Of DUI-DWI
Sent: Wednesday, August 11, 2004 2:25 PM
To: spambayes at python.org
Subject: [Spambayes] Question about the Outlook add-in and theclassification
of already filtered messages.


I have a question regarding classification with the add-in in Outlook.  Say
I get in 10 new spammy messages in the junk folder, and I trained the 3
lowest scoring spam as spam.  Now I want to re-score the remaining 7 spammy
messages to see whether their score has increased either substancially or
into spam range.  I try to achieve this by filtering the folder by NOT
actually filtering and only scoring.  I've found that when I do this, the
score actually decreases a lot, and it should bump up the score, especially
if you trained on a low scoring spam and it has duplicates (should be close
to 100).  The result is that they score between 3%-30% and should be the
other way around at 90%-100%.  Is there something that I'm doing wrong here?
 
To sum it up, I'm asking if I can classify messages either by single message
or whole folders with the Outlook add-in.  I know that you can classify
single messages in the proxy version.  Is there a way to un-filter messages
so that you can re-filter them by only scoring to get the proper scores?
 
Thanks!
 
Webmaster
 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.python.org/pipermail/spambayes/attachments/20040811/387c21c8/attachment.html