Oversight in MailScanner's Bayes Implementation?

[ISP List]

At 08:41 PM 12/13/2003 +0000, Ryan D. Egeland wrote:
>It appears the Bayes feature available through spamassassin specifically
>the way MailScanner implements it evaluates all incoming mail in a bulk
>fashion, i.e. each individual user does not have his own Bayes database.
>
>Is my assumption correct?

Yes.

>If so, it seems the power of the Bayes analysis seems markedly reduced
>through the default MailScanner configuration.

Somewhat, maybe.  However, there is no practical/scalable way to implement
a per-user Bayes database that MailScanner (or rather, SA) checks that I
can think of.  The best approach is to use MailScanner+SA at the relays and
let the user do a second line of Bayesian filtering on their mail client
(many of which are adding Bayesian stuff, e.g. Mozilla Mail/Thunderbird,
Eudora 6.0, SpamBayes for Outlook, and others..)


---------------------------------------
Mike Bacher / mike at sparklogic.com
SparkLogic Development / ISP Consulting
Use OptiGold ISP? Check out OptiSkin!
http://www.sparklogic.com/optiskin/
---------------------------------------