Mon, Apr 19, 2004 at 04:26:50PM +0200, Philippe Gramoull? wrote: > > Hello, > > Maybe you should give a try at crm114 ( http://crm114.sf.net ) by Bill Yera= > zunis Ph.D > > It's using Markovian filters instead of Bayesian ones and the homepage clai= > ms > 1 spam out of 8738 messages (4240 spam, 4498 nonspam) which translates into= > an > accurracy of 99.984% > > One nice aspect is that it learns *fast*
sound like pretty sharp accuracy. i'll take a look :) -- "the liberation loophole will make it clear.." lex lyamin
