ISSUE-74: MappingPropertyConventions

ISSUE-74: MappingPropertyConventions

Raised by: Alistair Miles
On product: SKOS

Currently, given separate vocabularies for semantic relations (skos:broader,
skos:narrower, skos:related etc.) and for concept mapping relations
(skos:broadMatch, skos:narrowMatch, skos:relatedMatch etc.), our assumption is
that, *by convention*, the SKOS semantic relation properties are *only* used to
state links between conceptual resources within the *same* concept scheme, and
the SKOS concept mapping properties are *only* used to state links between
conceptual resources in *different* concept schemes. 

Is this usage convention appropriate, useful and viable? Are there circumstances
where we would recommend *not* adhering to this convention?

Note that it is difficult to formally state any integrity conditions which could
be used to enforce this usage convention. No such integrity conditions are
currently stated in the SKOS reference. Therefore, graphs such as:

<A> skos:broader <B> ; skos:related <C> .
<A> skos:inScheme <MyScheme> .
<B> skos:inScheme <AnotherScheme> .
<C> skos:inScheme <AnotherScheme> .
<MyScheme> owl:differentFrom <AnotherScheme> . 

... and:

<A> skos:broadMatch <B> ; skos:relatedMatch <C> .
<A> skos:inScheme <MyScheme> .
<B> skos:inScheme <MyScheme> .
<C> skos:inScheme <MyScheme> .

... are both consistent with the SKOS data model. If we agree on the usage
convention, can we live without any supporting integrity conditions?

Note finally that if we chose *not* to have a separate vocabulary for concept
mapping relations and for semantic relations, and we use only semantic relations
to assert links between conceptual resources, this issue goes away. I.e. the
resolution of this issue is dependant on the resolution of ISSUE-71.

Received on Wednesday, 16 January 2008 21:23:56 UTC