[clug] Spambayes

[David Gibson]

On Thu, May 22, 2003 at 12:29:33PM +1000, Nemo -earth native- wrote:
> On Wed, May 21, 2003 at 12:15:19PM +1000, David Gibson did utter:
> > > 
> > > Personally, I'd be interested to see a comparison of the relative
> > > effectiveness of bogofilter, spambayes, spamassassin, etc... for
> > > example, train each system on the same block of spam and ham, and then
> > > test each on a new block of email known to contain both ham and spam... 
> 
> I'm increasingly finding I'm putting thought into this... SO, straw poll
> time. 
> 
> Q: What spamfiltering system do people use?
> 
> Q: Is it trainable?
> 
> Q: If it is, do you update it with ALL messages once classified as
> spam/ham, or only with false pos/neg results? (or not at all? Or
> something else?)

This is actually something you want to be very careful about:
attempting to train any machine learning system with it's own output
is unlikely to work terribly well.

-- 
David Gibson			| For every complex problem there is a
david at gibson.dropbear.id.au	| solution which is simple, neat and
				| wrong.
http://www.ozlabs.org/people/dgibson