perm filename X.TEX[AM,DBL] blob sn#410722 filedate 1979-01-17 generic text, type C, neo UTF8
COMMENT āŠ—   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	\problem  AI application (28)
C00005 ENDMK
CāŠ—;
\problem  AI application (28)

While the issues raised by automation are important 
and difficult, few could dispute the goal of creating an AI program
to replace Howard Cosell.  This question deals with the design
of such a system, a computer program capable of following the action of 
a professional sports event, analyzing it, and reporting its analyses.
\bline
\noindent
(a) [5 points]  A priori, what aspects of a sport would you expect to make this
problem more/less difficult?  Based on this, select a sport (if you
truly are unfamiliar with all professional sports, you may choose
the task of commentator for a chess match).   Justify your choice.
\bline
\noindent
(b) [11 points] What kinds of information could be utilized by such a program,
to enable -- or merely facilitate -- its operation?  For each type,
indicate an appropriate representation (and/or data structure), a
rough estimate of the amount of information desirable, the difficulty
of obtaining it, and its value to the program.
\bline
\noindent
(c) [8 points]  Sketch the flow of control through the program.  Note how 
each type of knowledge mentioned in (b) above is accessed and used.
\bline
\noindent
(d) [4 points]  What are the pros and cons of taking such a knowledge-based 
approach?  Consider, e.g., changing the program to another sport, time/space
costs, debugging the system initially, etc.