[Spambayes] Feature Request

[Katz, Amir]

 Skip,
W/o diving into this interesting flame (:-) I would like to make one point
regarding your own setup - Myself, and I assume most other users of SB do
NOT have access to the step in mail food chain where procmail and similar
tools can be installed & configured. We're stuck with the organizational
policy of using an end-user mail client (like Outlook) so the only weapon
that we have is a plug-in in Outlook (or the SB proxy for POP3).

Just my $0.02 Euros.

Amir 

-----Original Message-----
From: Skip Montanaro [mailto:skip at pobox.com] 
Sent: Tuesday, April 19, 2005 19:37
To: Bryan D. Andrews
Cc: spambayes at python.org
Subject: RE: [Spambayes] Feature Request


    Bryan> Better yet if there was a whitelist that allows you to do some
    Bryan> sort of weighted scoring based on a percentage that could make it
    Bryan> even more effective.

That's pretty much what SpamBayes does already.  Who an email is from is one
of the clues in a message, but it is just one of many clues, not a hammer
used to bludgeon the classifier into submission.  Your full name scored 0.16
for me on the message I'm replying to.  In addition, your domain
(trendcore.com) as well as information in various received headers also
scored 0.16.  There are thus several clues in your mail message telling me
you're a good guy:

    from:addr:bandrews: 0.16;
    from:addr:trendcore.com: 0.16;
    from:name:bryan d. andrews: 0.16;
    message-id:@tatl0s10.trendinfluence.com: 0.16;
    received:209.159: 0.16;
    received:209.159.56: 0.16;
    received:209.159.56.50: 0.16;
    received:tatl0s10.trendinfluence.com: 0.16;
    received:trendinfluence.com: 0.16;

If someone was to send me spam and spoof your email address, it's likely
most of those hammy clues would be missing (probably all but the first two
above).

Coming back to the whitelist issue, why is the filter capability in Outlook,
Eudora, Mozilla Mail, procmail or other mail software not up to the task?
Even Outlook Express, as feeble as it is, has filtering capability modestly
better than a simple whitelist based on the From: field of a message.  I
happen to use procmail as my local delivery tool.  I define five delivery
recipes before SpamBayes gets a chance to look at a message.  They are used
to vector various types of messages off to specific mailboxes.  You should
be able to do the same in other mail software and probably do it more
flexibly than anything we would implement for SpamBayes.  In my case, three
of the five recipes catch viruses and two deliver emails with specific
subjects to appropriate mailboxes.  None would be amenable to handling by a
simple whitelist facility within SpamBayes.

Skip
_______________________________________________
Spambayes at python.org
http://mail.python.org/mailman/listinfo/spambayes
Check the FAQ before asking: http://spambayes.sf.net/faq.html