From: Micheal E. J. <mic...@gm...> - 2006-04-24 15:41:10
|
On 4/24/06, David A. Lee <dl...@ca...> wrote: > > I'm not trying to be critical of ASSP, I love it ! However ... spamprobe > is truely amazing. It has taken the basic Bayesian concept and tuned it > over 5 years of greuling experience and active development. In particul= ar, > I find it significantly better at avoiding false spam. Out of over 100,0= 00 > spams I have only ONCE had it pass through a non-spam as spam, even duri= ng > initial training, whereas ASSP has passed through dozens in only about 10= 00 > emails so far. I understand this is out-of-scope of the current ASSP > project and thats fine. I was just wondering if it might be architectua= lly > easy to plug in ... and if so I'd be willing to try it myself. However,= I > suspect it isnt, and thats fine too. What ASSP give me is a full site-w= ide > spam solution, which is way worth the tradeoff. > > I'm intrigued! I'll have to give it a look. Perhaps there are methodologies worth looking into as well. |