RE: Bayesian Mailet

[Carl Vorster <cc...@propartners.co.za>]

Hi,

No feedback for me?

Thanks

Carl

-----Original Message-----
From: Carl Vorster [mailto:ccnews@propartners.co.za] 
Sent: Wednesday, July 30, 2008 9:38 PM
To: 'James Users List'
Subject: Bayesian Mailet

Hi All,

I've got a couple of questions regarding the Bayesian Analysis Mailet and
process.

I've configured the BayesianAnalysisFeeder for ham and spam as well as the
BayesianAnalysis mailet.
I can feed ham and spam and I see data in the db tables.

The analysis mailet is also working as I can see mails in deadletter/spam.

My questions are:
1. What's the process being followed with the spam in deadletter/spam, do
you check in on a regular basis for mails that's being flagged as spam
that's not, or just delete everything on a regular basis, push it to null?

2. How do you get mail that's in deadletter/spam back to the user's inbox if
it's valid mail. Do you delete all the valid spam and use FromRepository to
get it out?

3. Can I create a spam mailbox for each user and push the spam into that
mailbox for them to browse if they want to; what's the best way to do this,
using a Forward mailet in the spam processor; is this a good idea at all?

4. When you feed ham and spam, you'll feed all mails that you receive and
wasn't picked up by the BayesianAnalysis mailet as spam to the spam feeder,
but how many and how often do you feed valid mails to the ham feeder, do you
initially feed all valid mails, or only some - are there best practices I
can follow here?

Any other tips, tricks and traps that I haven't thought off yet or ran into
will also be greatly appreciated.

Thanks

Carl








---------------------------------------------------------------------
To unsubscribe, e-mail: server-user-unsubscribe@james.apache.org
For additional commands, e-mail: server-user-help@james.apache.org