From: project2501@project2501.cor   
      
   thanks - i will look at what you say about the weights and think about it.   
      
   ps - i agree - all this is not fuzzy, as you say. the fuzzy part of the   
   process is in the contruction of the features. that is for   
   	x = {   
   	    feature1 : A/0.2 + B/0.5 + C/0.2   
   	    feature2 : D/0.8 + E/0.4   
   	    }   
   the memberships associated with A,B,C,D and E are derived from a fuzzy set   
   over the feature1 linguistic variable. and THIS is the stage at which the   
   domain knowledge is most crucial. the clustering process itself should be   
   less dependent on the domain being mondelled. that is, the shapes of the   
   fuzzy sets over feature_x are domain dependent.   
      
   again - thanks everyone for their very useful input - very grateful!   
      
   if/when i get this stuff done - i'll post a link so people can have a look!   
      
   p   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)