[Spambayes] setup training & filtering set

Erik Sargent esargent at americanconsulting.com
Mon Aug 11 01:11:32 EDT 2003


Mark,
I just started using the plug-in this week, so my memory is pretty fresh.

I was also confused that it wasn't turned on at install.

I would suggest that you set a default of Monitoring to the INBOX folder and
turn it on by default.

I would also recommend that you automatically add a "Junk E-mail" and "Junk
Suspects" folder during install and use those as the default destination for
flagged messages. If you get a copy of Outlook 2003 beta2 Technical Refresh,
Microsoft automatically adds a "Junk E-Mail" folder. Even better, they allow
you to right-click the folder and delete spam directly from that folder.
This would be a cool dovetail of features with 2003.

As for training, I wouldn't mind starting from zero, but it does seem like
you could start with a small database of popular and unique messages.

Thank you for your work on this - very cool. I was happy to read about it
and even happier that it seems to work!

Erik

-----Original Message-----
From: spambayes-bounces at python.org [mailto:spambayes-bounces at python.org] On
Behalf Of Mark Hammond
Sent: Sunday, August 10, 2003 3:50 PM
To: 'Coe, Bob'; spambayes at python.org
Subject: RE: [Spambayes] spambayesaddin plugin not enabled

> I'm going to make a radical suggestion. The issue raised by
> Ms Delaval arises daily in this forum, and I think it stems
> from the fact that Spambayes is not enabled by default (since
> you have to pick your way through the training process
> first). Why not make the initial training optional and enable
> Spambayes by default? You'd need to include a few sample ham
> and spam messages to prime the process, but that's easy
> enough. Doing it this way would retard Spambayes's ability to
> adapt itself to the individual user, but it would catch up
> eventually. And a user who actually bothered to read the
> installation instructions could do it right, just like now. ;^)

Everytime this comes up, someone points out that the many clues in such a
pre-built database would be subtly different - eg, the "to" address, the
real name, the name of your ISP in the headers, etc.

However, InBoxer does exactly this - I wonder what their experiences are.
Sean?

Other options would be:
* A "wizard" style interface for initial training.
* Auto-enabling SpamBayes once it is trained and configured.

Mark.





More information about the Spambayes mailing list