Re: Problems with Bayesian filtering

> 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.

> What is the solution to this little trick?

For the white-on-white or 1pt text, the white-on-white tag, or the 1pt
tag, will become Bayseian tokens themselves with a very high
spammishness weight.

Actually, I feel the next step beyond Bayesian filters may be
genetically evolved spam-recognition code.  It would need to be p-code,
rather than native machine code, both to keep "bugs" in the code (which
will occur during mutation) from bashing something they shouldn't and
to make the resulting filters portable between machines.

Of course, I say that without having tried anything of the sort myself
(though I've been very tempted), so I could well be totally wrong.

/~\ The ASCII				der Mouse
\ / Ribbon Campaign
 X  Against HTML	       [email protected]
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B