[Spambayes] Correct procedure for handling bad mail from legitimatelist?

[Jesse Pelton]

You can delete the messages without training on them, but I'd treat them
like any other spam. SpamBayes trains on all the tokens in the message,
not just the sender. It keeps track of how often a given token appears
in ham and spam, so if you train on these messages, it will learn that
messages from these lists are sometimes spam. If you haven't trained on
any messages from these lists (as ham) before, future messages may be
seen as spam, but training on a few more messages should correct that.
 
One advantage to this approach is that you'll be training on whatever
other spam clues there are in those messages, which should improve
scoring of future messages. Another is that you don't have to try to
work out what to do with each messages: SpamBayes is pretty clever in
its simple-minded way, and second-guessing it seldom seems to work out
well.

________________________________

From: spambayes-bounces at python.org [mailto:spambayes-bounces at python.org]
On Behalf Of Steven Healey
Sent: Monday, June 19, 2006 10:19 AM
To: 'spambayes at python.org'
Subject: [Spambayes] Correct procedure for handling bad mail from
legitimatelist?



OK, some mailing lists I subscribe to were apparently hit by a clever
spammer over the weekend.  The result is that I now have 20-30 mails
containing spam, but received from legitimate lists, in my Junk Suspects
directory.

What is the correct procedure for handling these?   I fear that if I
select Delete as Spam, the signature of the legitimate list will be
added to the database as a trigger.  But if I select Recover from Spam,
the spam signatures which caused the messages to be flagged will be
deleted from the database which I also don't want.


Can I just Delete those messages? 

Thanks. 

sPh 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.python.org/pipermail/spambayes/attachments/20060619/271a6294/attachment.html