[Spambayes] More on Training Disparity Issues

[Adam Walker]

On Jul 17, 2004, at 5:46 PM, Richard B Barger ABC APR wrote:
>
> Are you just trying to save me steps (otherwise I'd be training on 
> >each<
> message), or does training on the mbox file pass additional header
> information -- some of the hidden stuff -- along to SpamBayes?
>

I haven't looked into the formats closely, but I think the message 
source is very close to mbox format (mbox may have some added status 
headers -- which would be ignored -- and msg separators ). It is just 
quicker to create a folder, copy a bunch of messages to it and train on 
the folder.