Re: [SKOS] proposed resolution for Issue 4 - BroaderNarrowerSemantics

Afterthought about naming:

broaderTransitive => broaderAncestor
narrowerTransitive => narrowerDescendant
This assumes readers understand the intended difference (a la XPath) between parent/ancestor and child/descendant.

Of course, these names need also to be evaluated in terms of the overall discussion about the broader/narrower names (I have to think every time I use them; for me the current meaning in counterintuitive).


PS I also suggest this naming discussion should not hold up publication of new WD.

Guus Schreiber wrote:
> All,
> Based on the post-telecon discussion of today, I'm proposing the 
> resolution below.
> Guus
> Proposed RESOLUTION for Issue-44 BroiaderNarrowerSemantics
> 1. Given that:
> - in the thesaurus world BT (BroaderTerm) is used to assert a concept
> as the *direct* parent in of another concept in a BT/NT hierarchy, and
> - we prefer the skos:broader to resemble as closely as possible the
> intended meaning and/or accepted practice of using BT
> we RESOLVE that skos:broader is *not* a transitive property.
> 2. Given that:
> - in the thesaurus world there is  clear need to be able to talk about
> the transitive closure of the BT relation, and
> - it would be confusing if SKOS (given its RDF/OWL baseline) would
> *not* predefine such a transitive interpretation of BT, and
> - for reasons of logical consistency (see note below) a transitive 
> version of BT
> necessarily needs to be defined as a *superproperty* of its
> direct-parent variant
> we RESOLVE that the property skos:broaderTransitive is added to the
> SKOS vocabulary, and that this new property is defined as a
> owl:Transitive property and as a superproperty of skos:broader.
> 3. Identical resolutions hold for the inverse properties skos:narrower
> and skos:narrowerTransitive.
> NOTE: The extension of a transitive variant B of a property A always
> contains the same or more instances than the extension of property A.
> Therefore, B cannot be defined as a subproperty of A (unless A and B
> are the same, which is not what we intend when introducing B).  So,
> transitivity is *not* automatically inherited over the subproperty
> relationship. For a description of a similar modeling pattern, see
> the SWBPD note on representing simple part-whole relations [1].
> [1]

VU University Amsterdam, Computer Science
De Boelelaan 1081a, 1081 HV Amsterdam, The Netherlands
T: +31 20 598 7739/7718; F: +31 84 712 1446 
Home page:

Received on Wednesday, 16 January 2008 01:21:58 UTC