– DRAFT –
RDF-star Semantics TF

<doerthe> I leave for a moment, I will be right back

<AndyS> From 2024-08-23:

<AndyS> > STRAWPOLL: Do you support the idea that in RDF semantics only predicates that are instances of rdf:ReificationProperty can have triple terms as objects?

<AndyS> > STRAWPOLL: Any IRI used as the predicate of a triple whose object is a triple term denotes an instance/\ of the denotation of rdf:ReificationProperty.

<niklasl> This https://github.com/w3c/rdf-star-wg/wiki/RDF-star-%22alternative-baseline%22

<doerthe> in my opinion, only the two extremes would work: the named triple and or the rather free properties

<niklasl> I'm inclined to agree. (Although I'm afraid of getting rid of "the lid on the can of triple terms" that restriction to rdf:reifies effectively is...)

<doerthe> the moment reifies is a predicate, we can declare an equivalent predicate and that is why I am against the restriction

<niklasl> There was a comment in a github issue thread which resonated with me: w3c/rdf-concepts#80 It compared using "bare" (my wording) triple terms with the regular use of "just" literals and "reified" literals (used in the comment to refer to the loose convention of rdf:value).

<gb> CLOSED Issue 80 where are triple terms allowed (by pfps) [spec:substantive]

<AndyS> Enrico's "alternative-baseline" defn for rdf:ReificationProperty -- https://github.com/w3c/rdf-star-wg/wiki/RDF-star-%22alternative-baseline%22#metamodelling-entailment-patterns-and-axiomatic-triples

<AndyS> Note 1: We could have "best practice" text to show how to use rdf:reifies.

<AndyS> Note 2: There seem to be only possibilities rdf:reifies must be used (and no others) or allow other properties and use the metamodel proposal for rdf:ReificationProperty.

<AndyS> Note 3: RDF Concepts (https://www.w3.org/TR/rdf12-concepts/#section-triples) currently has text to say "Every triple with a triple term as its object SHOULD use rdf:reifies as its predicate."

<AndyS> and also "Every triple whose object is not a triple term SHOULD NOT use rdf:reifies as its predicate."

<niklasl> foaf:Person owl:disjointWith rdf:TripleTerm .

<niklasl> <r1> rdf:reifies <bob> . <bob> a foaf:Person .

<doerthe> yes, but we are in RDF

<niklasl> Yes; I meant domain contradiction.

<niklasl> "An application might be upset" level. ;)

<niklasl> (Or a person ;P )

<doerthe> ys, I was just surprised because if what we add allows to construct contradiction in RDF itself, then RDF gets stronger and then we need to be careful

<niklasl> Ah, yes.

<doerthe> OK, I like domain contradiction vs. logical contradiction, we get a domain contradiction

<AndyS> Note 2: There seem to be only 2 possibilities (1) rdf:reifies must be used (and no others) or (2) allow other properties and use the metamodel proposal for rdf:ReificationProperty.

<doerthe> the one instance could be problematic

<AndyS> Or for 2.1 -- class rdf:ReificationProperty only has one member : rdf:reifies. Allows the future possibility of other properties. It is a fixed set and it is defined by RDF specs.

<niklasl> Or an application can say (in "its" restrictive domain model): rdf:ReificationProperty owl:oneOf (rdf:reifies) . # Not sure it that's good advice, and where to put it if so... (It's OWL, so not in any RDF docs...)

<AndyS> ... this leads to problems via "equivalence" via RDFS or OWL.

<TallTed> what is "the metamodel proposal for rdf:ReificationProperty" (in Note 2)?

<niklasl> Yes (but that was sort of intentional. It's about what rdf:reifies *denotes*.)

<AndyS> https://github.com/w3c/rdf-star-wg/wiki/RDF-star-%22alternative-baseline%22#metamodelling-entailment-patterns-and-axiomatic-triples

<AndyS> I will transcribe the notes into a new wiki page.

<niklasl> So, a propoal doc in https://github.com/w3c/rdf-star-wg/tree/main/docs ?

<niklasl> I do think in any case we need: rdf:reifies rdfs:range rdf:TripleTerm . # But that this might not be enough.