COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002								B-419-E24
C00006 ENDMK
C⊗;
							B-419-E24

This paper is too brief to serve as a tutorial and too vague to constitute an
original research contribution.  It could well be that the author is making a
real contribution to "Hierarchical Reasoning" but this report does little to
convince the reader that he has done so.  Semantic nets are not new.  Others
(not referenced by the author) have reported on their design and use.  The basic
problems of programming Go have been extensively described.  We still lack
concrete results to show that someone has been to write a Go program that plays
a reasonable game.  If and when the author is able to do this he will have
material for a very good paper.

The author fails to distinguish clearly between work that has been reported
previously and the new contributions that he is now describing.  If the author
is reporting his use of semantic nets for the first time he should point out, in
detail, why he has found this approach of more value than his previous approach.
This information would be of interest to other workers who ar currently
poundering the usefullness of such nets in their own work. On the other hand, if
he had already reported on the use of semantic nets, as I am led to believe from
the dogmatic statement that "Rag represents Go images by means of a semantic
network,..", then the present paper should present some detailed results to
demonstrate his success in applying these nets to Go, results, I might add, that
the reader could weigh for himself in deciding on the value of the work
reported.

The author is quite vague as to the source of the nets shown in Figures 2, 4 and
5.  Were these nets actually produced by his computer program from the Go
Board situation shown in Figure 1?  The author does not say that they were and
the presumption is that they were not.

It is one thing to draw a few relational nets by hand, and it is quite another
thing to write a program that will derive these nets and that will then be able
to use them in arriving at a reasonable course of action in an actual game
situation.  If the author's program is, in fact, able to produce such nets and
to use them effectively then the details as to the actual methods used would be
extremely interesting and would make a valuable contribution to the field.