From: jbbrus@comcast.net   
      
   Mark Nicholls wrote:   
   > I'm investigating a Rete rules engine (microsoft business rules   
   > engine).   
   >   
   > My problem is that Rete is obviously a different processing paradigm to   
   > conventional software....   
   >   
   > Does anyone know of some documentation that takes you through a good   
   > approach to 'designing' rules......I'm at an embyonic, spaghetti rules   
   > stage, and want to jump into an embryonic structured rules stage..   
   >   
   >   
      
   The classic reference to the RETE data structures and algorithm is the   
   following:   
      
   \bibitem{forgy} Forgy, C.L., RETE: a fast algorithm for the many   
           pattern/many object pattern matching problem, {\em Artificial   
           Intelligence} {\bf 19} (1982) 17--37.   
      
   If I understand your message correctly, it seems that you believe that   
   RETE has something to do with the problem-solving that is apparent to   
   the developer of the application system. It doesn't. Rather it is an   
   internal optimization technique to speed up the determination of which   
   rules or knowledge sources can execute at "this time". It does this by   
   maintaining partial matches of the rules' patterns to the blackboard or   
   system memory. The trick is that when small changes are made to that   
   memory -- say by rule execution -- only a relatively small amount of   
   repair is necessary to the partial matches. In some cases, the RETE   
   algorithm can reduce matching time by factors of thousands or millions.   
   The algorithm is used in many non-AI applications too. For example, if   
   one wants to establish a set of guards of daemons to protect a database   
   from particular inconsistent updates.   
      
   -- Jeff Barnett   
      
   [ 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)