XPost: comp.ai.edu, comp.ai.shells   
   From: scheuring@lycos.com   
      
   man@acm.org (MKN) wrote in message news:<400e3d1f$1@news.unimelb.edu.au>...   
      
   > XML is a markup language. It does not have sufficient control   
   > structure to code an intelligent system.   
      
   XML/XSLT is proven to be Turing-complete [1], which means that you can   
   code any necessary control structure you might need with it. Some   
   people use an XML-type language called AIML (Artificial Intelligence   
   Markup Language) to write inference engines with properties not found   
   in any other existing engine/language, and refuse to be discounted   
   from the ranks of those who are coding intelligent systems.   
      
   > Please investigate   
   > CLIPS (NASA, C language) and Jess (Sandia, Java), then decide   
   > if you want to re-code what has been done, in XML.   
   >   
      
   My goal is the simulation of personality by a computer, and I've yet   
   to find a language that was designed for that purpose, so there /are/   
   cases in which starting from scratch cannot be avoided. It sure can be   
   inconvenient, since a language like AIML, though Turing-complete, only   
   provides the programmer with primitive operations. Once you've managed   
   to build higher-level, reusable objects from those primitives, though,   
   things start to get better.   
      
   > If your goal is a professional (as to research) product. I would   
   > recommend that you use an existing deduction engine and use   
   > XML as an inter-exchange format to the external world.   
   >   
      
   AIML can be (and is) used for interfacing with engines like OpenCyc   
   [2] and Prolog [3]. Though this is certainly useful, it doesn't help   
   any with the 'personality simulation' problem, since those systems   
   were designed /not/ to regard any of their knowledge as something   
   involving 'personality' (and for good reasons, too). So if you need a   
   front end with a 'virtual personality', it's actually your best bet to   
   code up a lightweight inference engine in XML, and let that talk to   
   Cyc only when necessary.   
      
   All the best,   
      
   Dirk   
      
   [1] http://www.google.com/search?q=xslt+turing+machine   
   [2] http://www.daxtron.com/pdf/Cyn_Description.pdf   
   [3] http://list.alicebot.org/pipermail/alicebot-style/2001-November/000041.html   
      
   [ comp.ai is moderated.  To submit, just post and be patient, or if ]   
   [ that fails mail your article to , and ]   
   [ ask your news administrator to fix the problems with your system. ]   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)