Unicity of the Couple: important for designing User Interface to manage SKOS triples

Hi!

While trying to design a good user interface for managing SKOS 
statements, I thought to identify the statements to update by themselves 
(i.e. by the RDF triple: subject=SKOS concept, predicate=SKOS property, 
object=label, note or other SKOS concept).
Each statement would become a single data unit to which an 
approval/deprecation workflow can be applied.

To modify the workflow status of a statement, the user interface was then:
1) choose a SKOS concept to update
2) get it displayed in "edit mode"
3) scroll down to the SKOS relation for which you want to update a statement
4) choose the statement with the right object linked (label, note, SKOS 
concept)
5) pop up a form to update the workflow status of that statement

Showing sketches of this to the users, first reaction was "how do I 
change a term from prefLabel to altLabel?"
Good question (I mean annoying one!)!

I would like to know if the SKOS community generally agrees that within 
the SKOS relations, not only the triples (subject-predicate-object) are 
unique, but also the couples (subject-object).

For instance, in practice:
* a given term for a Concept can be a prefLabel, an altLabel, an 
hiddenLabel but never two of them
* a given SKOS concept can be a broader, a narrower, a related, a 
closeMatch, an exactMatch, a relatedMatch, a narrowMatch, a broadMatch 
but never two of those
Are all relations disjoint from the others IN PRACTICE?

If this is true in your practice (it is not in the current formal 
definition), then it simplifies greatly the management of the data for 
the users: the type of label (pref, alt, hidden) or the kind of relation 
becomes (from the user point of view only) an attribute of a unique 
couple (subject-object). The workflow then allows to approve/change 
property/deprecate any couple (instead of approve/deprecate of triples): 
this make life a lot easier for users.

The current SKOS reference is already stating that <A> broader <A> is 
theoretically allowed but hardly accepted by most management software. 
Should we state somehow that "duplicated subject-object" relations are 
permitted but are disjoint in practice?

Reminder: I am using SKOS for thesauri, indexing vocabularies, authority 
lists NOT for "reasoning" on ontologies.

Thank you for your insights!!!

Christophe

Received on Wednesday, 6 July 2011 17:06:39 UTC