Training with mailbox format files

[David Relson]

At 06:42 PM 7/31/03, Rob Tanner wrote:
>Hi,
>
>We've been using bogofilter now for several weeks and since we're still
>fine tuning, we deliver all mail, SPAM or not -- but prepend [SPAM] to the
>subject header when it's SPAM.  I have a couple mailboxes that folks copy
>misclassified messages into, and I use those to further train bogofilter.
>
>The easiest way for me to collect the messages is to drag them from the
>server to a local mailbox on my workstation.  The problem is that the
>software I use, Mulberry, writes them to a single rfc822 format mailbox
>file.  Can I train bogofilter by giving it that single file, or must I
>break the messages up and feed them one at a time?
>
>Thanks,
>Rob

Hi Rob,

Bogofilter has understood mailbox files from the beginning.  Use 
"bogofilter -s < spam.mbx" or "bogofilter -n < ham.mbx" as appropriate.  In 
you include the "-v" switch, it will tell you how many words and messages 
were processed.

David