There is another reason to use SKOS as a view for Class models (which a form of polymorphism known as OWL "punning" apparently)
SKOS preserves direct parent/child inheritance irrespective of the entailment regime. RDFS subClassOf does not allow you to identify the immediate parent, and this is not relevant to the set theory aspect of reasoning. Actual models have explicit semantic intent - as evidenced by the desire to build class heirarchies. If you fully entail RDFS everying is a subclass of itself and every other ancestor in tree. I feel that SKOS is playing the role of a "shape" for conceptual hierarchies that may be further described in OWL .. On Wednesday, 14 April 2021 at 21:06:08 UTC+10 David Price wrote: > Hi TB users, > > Just thought I’d add that I am not suggesting that using a reasoner is of > primary importance to most LD/Semantic user scenarios. > > I’m really trying to understand why a specific ontology (SKOS) is being > used for something that seems to be included directly in both the OWL and > SHACL languages out-of-the-box. > > Cheers, > David > > > On 14 Apr 2021, at 10:24, David Price <[email protected]> wrote: > > Hi TB users, > > The discussion with Michel caused a question to pop into my mind and I > hope to get some customer feedback here. > > I see customers/bodies use SKOS to define what turn out to be simple lists > of standard class instances that are then used in an ontology. My question > is - Why? > > OWL enumerated classes provide the same standard-instance capability, but > without the added “clutter” in your ontology when you import SKOS. The > number of completely empty, never-to-be-used cases of Collection, Concept > Scheme, etc. appearing as classes in ontologies has become surprising to me. > > Can anyone who’s been part of making this kind of decision in your > ontologies please explain Why? Among other possible reasons, a few I could > think of might be: > > - Is it that the UI for creating them is simpler? > - Is it that these standard instances are seen as being somehow > improved by having a not-well-defined tree structure available via > broader/narrower? > - Is it that you’re using a standard that’s done this and you’d be > happy with any solution that supports defining standard instances? i.e. > perhaps a SKOS-to-enumerated-class importer for these situations would be > a > useful tool? > - Is that you really have use cases like described here: > https://www.w3.org/2006/07/SWD/SKOS/skos-and-owl/master.html. ? > - Is it really just that you prefer/need the SKOS annotation > properties, even for classes, instances, etc? (I like them and have local > a > skos-annotations-only.ttl I reuse) > > We are always looking for real-world user scenarios wrt improving EDG, and > so you can also think of this as a requirements gathering/feature request > exercise. > > It may be there are some EDG feature or UI changes we could make. It may > be that a How-To video on our Web site explaining this approach and how to > best execute it in EDG would be useful. I don’t know but thought it worth > asking the requirements question to see how prevalent this is and why it > happens. > > Cheers, > David > > > “ > *x.y Combination of language bindings* > > In practice, there will often be a need for a combination of language > bindings. In the most extensive case, this concerns (always 1 or more): > > - SKOS concept scheme; > - RDFS (basic) ontology (imports the SKOS concept scheme); > - OWL ontology (imports the RDFS ontology); > - SHACL graph (imports the RDFS ontology). > > Since the semantic intent of the SKOS concepts is fundamentally different > from that of the ontological resources describing the semantics of the data > and/or serving as the basis for reasoning about data, it is important to > distinguish the name space URIs from the SKOS concepts of the name space > URIs of the ontology resources by using a different name space. The > ontological items are linked to the corresponding SKOS items using the > existing rdfs:isDefinedBy relationship. > > Since RDFS, OWL and SHACL are not fundamentally different and cover > complementary semantic aspects, the same name space URIs can be used > mutually for resources defined 1) with RDFS only, 2) with RDFS plus OWL and > 3) with RDFS plus SHACL. A resource can be typed as an rdfs:Class, an > owl:Class and a sh:NodeShape at the same time. The ontologies represented > in RDFS, OWL and SHACL are recorded in their own graphs with their own > graph URI and can therefore be used independently of each other and import > each other. > > > > UK +44 (0) 7788 561308 <+44%207788%20561308> > US +1 (336) 283-0808 <(336)%20283-0808> > > > UK +44 (0) 7788 561308 <+44%207788%20561308> > US +1 (336) 283-0808 <(336)%20283-0808> > > -- You received this message because you are subscribed to the Google Groups "TopBraid Suite Users" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/topbraid-users/70fc5922-bd00-4295-8e8c-5e1cc241f977n%40googlegroups.com.
