Thanks to all who replied. This sums it up quite well and confirms my doubts about the algorithm-exclusive methods. As always I appreciate this list's objectivity on the topic.
OK, but if it is true that their "learnig methods" are better than SA's Bayes, than what about creating a SA plugin to use DSPAM as ONE of SA tests?
Bye.
--
___________________________________________________
__
|- [EMAIL PROTECTED]
|ederico Giannici http://www.neomedia.it
___________________________________________________
