XPost: comp.ai.nat-lang   
   From: mailbox@dmitry-kazakov.de   
      
   On Wed, 14 May 2008 10:20:06 GMT, amnon@textanalysis.com wrote:   
      
   > Computing with Evidence   
   >   
   > (Note to Dmitry -- I'm doing everything manually at this point.  You   
   > can add automated learning to the next draft!)   
   >   
   > While the latest thread on this topic was titled *Computing with   
   > Confidence*, I realize that what I'm really doing is computing with   
   > *evidence*.  The NLP++ confidence operator (%%) is perhaps more   
   > properly thought of as an evidence operator.   
   >   
   > So "my" theory is that one can accumulate an infinite amount of   
   > evidence for and against something, but the total amount of evidence   
   > of either kind never exceeds 100%.  Given 80% evidence for something   
   > and adding a new datum at 70% *evidence level* yields something like   
   >   
   > 80 %% 70 = 86%   
   >   
   > So my total accumulated evidence is now at the 86% level.   
      
   For evidence to be accumulated infinitely, they have to be in part   
   independent. If you do not add/remove anything new, there is no reason to   
   change the level (= the measure of the accumulated set of evidences), that   
   would be counterintuitive.   
      
   In short, %% obeying any reasonable set of axioms + having your   
   accumulation property cannot be a function of its arguments. It must be a   
   function of *evidences*, not levels (a measure of).   
      
   --   
   Regards,   
   Dmitry A. Kazakov   
   http://www.dmitry-kazakov.de   
      
   [ comp.ai is moderated ... your article may take a while to appear. ]   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)