From: grsNOSPAM@NO-SPAMmail.com   
      
   > Can you give as an idea, how do you formalize the concept TALL without   
   > taking its membership function off the wall?   
      
   It seems like in your particular example, you're not even taking the   
   time to think.   
      
   When you use fuzzy sets to map "concepts", you're defining them in   
   a specific context - the fuzzy system you're working on. It doesn't   
   even matter what each of them are called, as long as it's congruent with   
   the rules that depend on them.   
      
   Thus, the point is not to "formalize" specific concepts, but to give   
   them some meaning that's useful within a specific context.   
      
   In other words, I think you're trying to "fight" fuzzy logic with   
   the wrong ideas. How can a scientist ask polemical questions about   
   a subject they haven't cared to study and about which they only   
   have a few "popular misconceptions"?   
      
   I remember a conversation I once had with my logics professor back   
   when I was a MD student. We were talking about fuzzy logic, and he   
   basically told me that he didn't think there was a fundamental   
   difference between classical logic and fuzzy logic, and thus no   
   real benefit either. One of his points was that in real-world   
   applications, fuzzy systems are ultimately implemented on digital   
   processors that, at a low level, only execute plain logic operations.   
   So there was a formal way to transform any fuzzy logic system into a   
   regular logic, rule-based system. He wasn't totally wrong, but I think   
   he was missing the idea of what a "tool" is. You can probably set a fire   
   with a couple wood sticks (like in old times), but you'll get this   
   result very much quicker, and more consistently, with a lighter.   
   The same goes with fuzzy logic: there are a log of areas where it can   
   help you get the results you're aiming for quicker, and in a more   
   consistent manner.   
      
   Elaborating on this "fuzzy logic --> classical logic" idea, there   
   are of course theoretical cases where you can't map a fuzzy logic   
   system into a classical logic one. I'm especially thinking of   
   continuous domains (vs. discrete). Up to now, fuzzy logic applications   
   usually involve discrete domains, but that may not always be the case.   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)