[sc34wg3] Association items

Nikita Ogievetsky sc34wg3@isotopicmaps.org
Sun, 12 Jun 2005 23:00:18 -0400


! * Lars Marius Garshol
! |
! | That's pretty clear from the prose: the intended semantics is that of
! | a relationship.
! * Nikita Ogievetsky
! |
! | But it is not clear from the historical use of associations,
! | especially the way that they were meant to replace facets.
! Well, they weren't meant to replace facets; they were in topic maps
! before facets were. It was just realized that when topics can
! represent information resources, the reach of associations is extended
! to areas previously only covered by facets.

Hmmm... It was always possible to represent this with occurrences... or I
missed something...

! Therefore, facets were no
! longer needed. (And, in any case, facets *did* represent
! relationships; their purpose was to represent resource -> topic
! relationships.)

"The facet value (fvalue) element form specifies a user-defined value of the
property (facet type) being applied by the containing facet link. The
information objects to which the property/value pair is being assigned are
referenced by means of the location addresses specified in the content of
the fvalue element." [1]

To me facets where meant to indicate properties like age, weight, height,
etc. (monadic predicates indeed). Now most people use occurrences for this
(as far as I know - not associations as you and Murray suggest - please
correct me if I am wrong). Use of occurrences for this purpose is quite
weird, because occurrences were originally meant to represent relationships
between topics and resources (occurrence of a topic in a resource).  
! | I wonder if facets should be considered to be brought back into XTM.
! For the reason given above (there's nothing they can do that cannot be
! done with topics and associations) I don't think there's any point in
! doing that.
! | On the other hand, even "monadic" associations (re: Murray's posts
! | on the other sub-thread) have a few topics related - if you count
! | association type and scope.
! True, but I don't think it is right to count these. They define the
! nature of the relationship, and don't count as being among the things
! related.

Agreed. However, I am not sure what is right and what is wrong. In the
current model(s) they perhaps are not counted, but actually they all are
related through an association. On the other hand, topic players themselves
help define the nature of association (extensive definition). 


[1] http://www.y12.doe.gov/sgml/sc34/document/0085TXT.pdf