[Spambayes] Incremental Training

[Mike Iles]

I've been running SpamBayes for a few weeks now and find it to be
extremely accurate, even with a relatively low training volume (500
good, 80 spam). Thanks for putting out such a great tool!

 

My question is about the incremental training: it appears to me that
when SpamBayes classifies a message as spam it doesn't also train
against it. I would think that this would be a very nice attribute: if
the spam I was receiving was evolving in a given direction SpamBayes
would then track it without my intervention.

 

I realize that I can continue to train by occasionally unchecking the
'rebuild database' box and clicking the 'train' button again. Happily,
this seems to recognize the duplicates and doesn't train twice against
them.

 

In any case, am I missing a configuration option to enable this? If not,
is the method I've described above the recommended way to continue to
train against the spam I'm receiving?

 

Many thanks,

Mike.