[topicmapmail] generate hierarchical GUIs from all transitive associations

[Lars Marius Garshol]

* Conal Tuohy
| 
| Yes I think that's the author's idea - and the occurrence type is
| another topic in the same topic map: "association-property", which
| represents these "set-theoretic" properties (transitivity,
| reflexivity, etc).

>From looking at the definition of association-property it seems that
you are right.
 
* Lars Marius Garshol
|
| That's hardly recommended practice any more, but probably made sense
| back in March 2001 when this was written.
 
* Conal Tuohy
|
| Tell me more - how would you recommend marking up this particular
| feature of the association, Lars?

Clearly 

  http://www.topicmaps.com/xtm/1.0/template.xtm#assoc-prop-transitive

is a published subject identifier, not an occurrence. That means there
should be a topic that has this as its subject identifier, and this
should be related to the assocation type by some association. One
alternative might be a has-property association type. Some people
might also use it as the class of the association type, though I'd be
more wary of that.

In Ontopia we were very enthusiastic about creating PSIs for various
kinds of properties association types might have, before we realized
that inference rules do the same thing, only much more strongly.

  greater-than($A, $B) :- {
    comparison($A : greater, $B : smaller) 
  |
    comparison($A : greater, $C : smaller), greater-than($C, $B)
  }.

This recursive rule implements the transitivity of the
comparison-association. 

  uncle-of($U, $N) :-
    parent-of($G : parent, $U : child),
    parent-of($G : parent, $P : child),
    parent-of($P : parent, $N : child),
    male($U).

This rule goes way beyond anything classifying associations as
transitive, reflexive, etc could ever hope to do.

Once this became clear we felt working on inferencing was more
important and have done so since, though it is not entirely clear to
me that inferencing makes annotation of association types entirely
superfluous, or whether there is room for both.

In any case a proper analysis of the various possibilities and how
they play out in topic maps would be very interesting. Would be great
if someone could take the time to pick up this idea and write up the
analysis.

-- 
Lars Marius Garshol, Ontopian         <URL: http://www.ontopia.net >
GSM: +47 98 21 55 50                  <URL: http://www.garshol.priv.no >