[sc34wg3] TMDM draft - Comments

Lars Marius Garshol sc34wg3@isotopicmaps.org
Thu, 14 Jul 2005 21:13:45 +0200

Having said that, let me move on to your comments.

* Lars Heuer
| 5.7 Association items
|     The paragraph behind the example describes that all associations
|     have a scope but does not mention the unconstrained scope
|     explicitly as done in the section 5.3.4 "Scope" for the topic
|     characteristics.

5.3.4 defines the notion of scope and so has to define the
unconstrained scope. However, the unconstrained scope (and scope) can
only be defined once.=20
|     Maybe "Scope" should get a dedicated section and a description
|     where topic characteristics _and_ associations are mentioned;
|     otherwise the paragraph about the association's scope property
|     be updated with and name the unconstrained scope explicitly.

The definition cannot be made without a notion of topic
characteristics, which again cannot be defined without a notion of
topics. It *is* possible to move 5.3.4 out of 5.3, but this doesn't
seem like a good solution, nor does there seem to be any problem to
which this might be the solution.

(Sorry if I seem to be writing ISO-ese. I've done nothing else for the
past week, and doubtless it will take some time for the effects to
| 7.5. Subjects for defined terms
|      http://psi.topicmaps.com/iso13250/topic-characteristic
|      [...]
|      Usage: common supertype for association, occurrence, and topic name.
|                                  ^^^
|                                  association _role_

| 5.3.4 Scope
| [...]
| The unconstrained scope is the scope used to indicate that a topic
| characteristic is considered to have unlimited validity. In the model
| this is represented by the empty set.
| [...]
| 7.5 Subjects for defined terms
| [...]
| http://psi.topicmaps.com/iso13250/unconstrained-scope
| Definition: scope used to indicate that a topic characteristic is
| considered to have unlimited validity.
| [...]
| First: Associations are not mentioned here.

Correct. I am not sure if they should be. Input on this would be
| Does not the description in 5.3.4. contradict the definition in 7.5?

If it's considered that the scope really applies to the assignment of
roles to topics rather than the association itself: no.

| I think either in the model the unconstrained scope must be
| represented by a topic with the subject identifier
| "http://psi.topicmaps.com/iso13250/unconstrained-scope" or the
| subject's description should be changed to "scope property of a
| topic characteristic or association that is the empty set" or
| something like that.

If a topic for the unconstrained scope were created it would break the
property that adding a topic to a scope narrows its validity, and it
would also mean that if u is the unconstrained scope topic, then for
all scopes s (except s =3D {u}),

  s - u =3D s

which is very, very ugly indeed.

But I agree that this definition seems dangerous when used here. The
definition was harvested from the source of the document (as for all
the other PSIs), but it seems that in this particular case we need to
change it. I'll make this an issue for Montr=E9al; input welcome in the

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