Since already 4 months I'm using in production a James mailet implementation of Paul 
Graham's paper. See http://www.mailet.org/directory.html. It works greatly.

It's an extension from Paul Grahams' "A Plan for Spam" 
(http://paulgraham.com/spam.html) to almost all of his "Better Bayesian Filtering" 
paper(http://paulgraham.com/better.html) that I did to Chris Mean's mailets (see 
http://archives.apache.org/eyebrowse/ReadMsg?listId=21&msgNo=4481 and related thread), 
that he is using since an even longer time.

Vincenzo

> -----Original Message-----
> From: Sanne de Roever [mailto:[EMAIL PROTECTED]
> Sent: giovedì 24 luglio 2003 19.00
> To: [EMAIL PROTECTED]
> Subject: spam filter/ Message ID
> 
> 
> Hi,
> 
> Is there any work on integrating a spam filter in James?
> 
> Having read the expose by Paul Graham I thought of the following scheme.
> Spam messages could be forwarded to the James server to a mailet (spam@)
> which could move the spam messages to a spam mailbox based on the 
> message ID
> (is this always present?). Using classifier4j a classifier could then be
> trained to recognize these message from the messages in non spam boxes.
> 
> The idea is obvious, but is there any work being done on that?
> 
> Would it be a good idea?
> 
> Sanne
> 
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
> 


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to