First off, despite my e-mail address at the University, I am a staff   
   person, not a prof.   
      
   I have been interested in issues regarding sparse Karnaugh maps for   
   quite some time.   
      
   If you have less than the full number of data points, can you make a   
   best guess?   
      
   Can you predict what data should be gathered to get a better guess,   
   rather than random   
   gathering?   
      
   Lets say you have points 4 (0100) and 9 (1001) which are of one type   
   (A) and   
   points 3 (0011) and F (1111) that are another type (B).  You can then   
   calculate   
   the possible regions for type B by the intersection of NOT 4 (1XXX,   
   X0XX, XX1X, XXX1)   
   and NOT 9 (0XXX, X1XX, XX1X, XXX0).   
      
   You can get the possible regions for type A in a similar manner.   
   Obviously, It helps to   
    have a computer program do this.   
      
   Where the two types intersect, you can seek data.  There will be cases   
   where there   
   are more intersections at some points than others, so you can seek   
   those that are most   
   likely to improve the guess.   
      
   You can also detect regions of the Karnaugh map that are not covered   
   by the possible regions of some set of data points (not necessarily   
   those above), which is another   
   target for data gathering.   
      
   If you can use bits, like the bits in a Karnaugh map, to make a guess   
   at the logic involved   
   between similar and dissimilar data points, it seems to me that it may   
   have some application to AI.  We may make such guesses all the time   
   based on insufficient data.   
      
   [ 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)