Re: Problems with Bayesian filtering

Carl Hutzler wrote:
> Our bayesian filters are being worked around by spammers using "chaff 
> text" which is either in white font on white background, 1 pt font, or 
> in some cases, just normal text a few CR's below the spammer's VERY 
> SHORT message.

In my implementation of Bayesian mail classification (POPFile) I 
specifically look for the use of trickery like white on white (I call it 
Invisible Ink), 1pt fonts etc. and use them as additional tokens for the 
classifier.   They are in fact very good indicators of spam.

As for the short message with random text at the bottom: this has not 
been a problem for me because the spammers are randomly hitting more 
spam words than ham words for me, but I could see this being a problem 
on a Bayesian filter used by a large organization because the ham 
database might be very large and wide ranging.

John.