Re: [tm-pubsubj-comment] ISSUE 4 - Relation of subjects; PSD structure

[Lars Marius Garshol <larsga@garshol.priv.no>]

* Mary Nishikawa
| 
| I have a few superclass-subclass relations defined as associations
| in my XTM TM. http://purl.org/mnishikawa/unspsc-core.xtm I needed to
| do this to have a realistic example, but I did not expect any
| description in our published subjects recommendations. This is
| really a modeling issue for TM. 

It is, but we do need to give people guidance on what kinds of
assertions about published subjects are recommended to make as part of
the formal definition of the subjects, which ones are better made
optional, and which ones are best omitted altogether.

I think this issue is at the same time thorny and crucial, and that we
must deal with it if what we do is to be of any use to anyone.

To take an example from GeoLang: is it acceptable for the PS set to
say as part of the core XTM file that "Mayotte" is an instance of the
class "country"? What if someone wants to refer to Mayotte, but they
disagree that it is a country, or perhaps they use their own concept
for the class?

| I was thinking that once the Pubsubj committee got the
| recommendations out, the GeoLang and XMLVoc would deal with this
| issue.

Those TCs will deal with the issue for the vocabularies they publish,
but dealing with the issue in the general case is way beyond their
scopes. Whether descriptions should be part of PSDs or not is really
an issue for the PubSubj TC to decide.

BTW: There are a number of issues with the unspsc-core.xtm file. Let
me know if you want to hear what they are (and where you'd like to
hear).

-- 
Lars Marius Garshol, Ontopian         <URL: http://www.ontopia.net >
ISO SC34/WG3, OASIS GeoLang TC        <URL: http://www.garshol.priv.no >