Bayesian filtering is ineffective at the MTA level. There are too many 
varying definitions of what constitutes spam - there will either be 
significant false positives, or significant false negatives.
I speak from experience here - I implemented an MTA level bayesian filter 
on my mail server last year. They only possible way to make it work was to 
have each user able to configure their own spam and ham databases - trying 
to do it globally just resulted in spam continuing to pour in (or if the 
algorithm was adjusted to prevent that, lots of bounced legitimate mail). 
And having individual user databases quickly becomes memory intensive, as 
these databases have to be scanned for every email received (not to mention 
having to come up with ways to allow the end user to update their own 
databases with new spam and ham samples over time).

This occurs because while Bob may feel that every marketing mail he 
receives is spam, Jane really wants to get her lottery newsletters and 
special offers from her local department stores, but doesn't want the 
viagra and porn spams, while Ralph really likes the porn spam, but doesn't 
want the rest.

At 08:53 2/10/2004, Edinilson J. Santos wrote:

>http://assp.sourceforge.net/
>
>
>Edinilson
>---------------------------------------------------------
>ATINET-Professional Web Hosting
>Tel Voz: (0xx11) 4412-0876
>http://www.atinet.com.br
>
>
>---
>Outgoing mail is certified Virus Free.
>Checked by AVG anti-virus system (http://www.grisoft.com).
>Version: 6.0.580 / Virus Database: 367 - Release Date: 06/02/2004
>-
>To unsubscribe from this list: send the line "unsubscribe xmail" in
>the body of a message to [EMAIL PROTECTED]
>For general help: send the line "help" in the body of a message to
>[EMAIL PROTECTED]


-
To unsubscribe from this list: send the line "unsubscribe xmail" in
the body of a message to [EMAIL PROTECTED]
For general help: send the line "help" in the body of a message to
[EMAIL PROTECTED]

Reply via email to