Meeting minutes
<niklasl> Treat rdf:reifies no more and no less specially than rdf:type?
<pfps> indeed, the question is what triple terms denote
<pfps> if they denote some arbitrary resourse then there is no "problem" with literals sitting in places where a triple term normally goes
<niklasl> So, what do triple terms denote?
<pfps> if, on the other hand, a triple term denotes some a element of some datatype then a literal in the object position is likely a contradiction
<niklasl> "<s> <p> <o>"^^rdf:TripleTerm 🤔
<pfps> The proposed semantics in PR #55 appears to make triple terms be arbitrary resources so there is no problem with literals in the place of triple terms.
<gb> Action 55 setup echidna in all GH repository (on pchampin) due 27 Apr 2023
<niklasl> An abstract proposition is a resource.
<niklasl> An IRI does not necessarily denote an IRI....
<pchampin> +1 to what niklasl said
<pchampin> the question is: what is the set of images of function RE
<tl> triple term are resources and denote propositions
<pfps> triple terms are syntactic elements, which are only incidentally resources
<pfps> just like IRIs are syntactic elements, and only incidentally resources
<niklasl> In the interpretation: "Is this proposition true?" => In the syntax: "Is this triple in the graph?"
<pfps> what is the difference between a triple term and an IRI here?
<william_vw> so will all Propositions be Resources as well?
<pfps> what is the difference between a triple term and a literal here?
<niklasl> I *think* range of rdf:reifies as rdf:Proposition makes sense (understanding that e.g. tl sees a potential for it to mean something wider). Just as rdf:type rdfs:range rdfs:Class .
<pfps> everything is a resource (well not exactly in simple semantics, but definitely in RDF semantics)
<niklasl> +1 to pchampin, they are all triples, used as terms and/or asserted.
<Zakim> pchampin, you wanted to talk abour rdf:Literal
<niklasl> Literals are almost as weird as named graphs. Literals are "a pair of their values and a datatype"... (I think it is written somewhere?)
<tl> i would rather treat triple terms as IRIs than as literals, because they do not denote "themselves" but something referenced by their constituents
<niklasl> From semantics: RDF literals combine a string and an IRI identifing a datatype.
<niklasl> I presume that refers to the syntactic structure though.
<pchampin> that's ok with me
<niklasl> Me too.
<pchampin> just like :john a "Person" means that "Person" is a class
<AndyS> so how to record this? "A triple term denotes ????" (text needed for the docs)
<pchampin> I missed the pattern too :)
<niklasl> "A triple terms denote a proposition". A resource in the class (with the rdf:type) rdfs:Proposition.
<tl> niklasl "A triple terms denote a proposition (not necessarily asserted in a graph)".
<niklasl> +1 for rdfs:Proposition (we have rdfs:Literal there; even rdfs:Resource). Some things in rdf: are defined using rdfs:.
<AndyS> ... and what is a proposition? (for the rdfs:label/comment!)
<pchampin> finally found it: the section explaining that "The class rdfs:Literal is not the class of literals, but rather that of literal values"
<niklasl> The property carries all the meaning if it is defined with a clear rdfs:range here. (I am more neutral about its domain.)
<Zakim> pfps, you wanted to comment on numeral vs number
<pchampin> rdf:string ??
<pfps> if there is an IRI for the value space of triple terms, then it belongs in the rdf namespace just like rdf:string
<niklasl> I think if we say triple terms denote (logical, "atomic", binary, directed) propositions, we are clear they denote "numbers not numerals"?
<pfps> well, a proposition is certainly different from a syntactic thing, so I would say yes
<niklasl> Just because a person is a resource and an apple is a resource, and I am a person, I am not an apple.
<pfps> that's not to say that I think a triple term should be a proposition, as I view propositions as "types", not "tokens"
<pfps> statement is something that can fit on either side, I think, and so is not a mistake in RDF reification
<niklasl> Unless in the interpretation model of a shark, who just consumes me. ;)
<tl> pfps but triple terms represent types, so what's the problem?
<pchampin> +1 pfps, and as triple terms are closer to "types" than "tokens", I think rdf(s):Proposition is a good fit
<pfps> hmm, I guess I agree with pchampin, that triple terms are closer to "types" because of their uniqueness
<pfps> this makes them different from statements, which are not unique
<tl> exactly
<pchampin> exactly, and a good way to explain the difference between triple terms and old-style reification :)
<tl> yes, but then enter the reifiers :->
<pfps> so I'm OK with using rdf:Proposition as the type for the denotion of triple terms, if such a type is needed
<pfps> that's not to say that there isn't a better name
<Zakim> pchampin, you wanted to suggest a rule for rdf: vs. rdfs:
<Souri> A naive question: Is RDF String a class or just a set of all the string values that can be specified in RDF syntax? What if RDF TripleTerm is (similarly?) just a set of all possible triple-terms allowed in the syntax. And rdf:reifies can only have elements of RDF TripleTerm (set) as its object. Would that allow us to avoid rdfs:domain and
<Souri> rdfs:range in RDF (which IMO adds complexity to RDF).
<niklasl> Aside: Is rdfs:Proposition a subclass of xsd:boolean (or vice versa)? (Mostly but not entirely joking.)
<pchampin> no no no no! 8-O
<pchampin> :)
<niklasl> No ;)
<pchampin> as pfps emphasized, a proposition is true or false *in a context*... so not a subclass of boolean
<william_vw> I'm a bit stunned at your interpretation of triple terms - e.g., they *cannot* be events - I hope that you won't expect people to know all these implicit restraints. They are not as intuitive as you may think ...
There is an agreeement that we have to define the range of rdf:reifies to be rdfs:Proposition
<enrico> +1 to leave the domain rdf:reifies unspecified
<niklasl> +1 to pchampin; rdfs:reifies rdfs:range rdfs:Propositon; rdfs:domain rdfs:Resource # Domain best left unspecified
<niklasl> That this is unintuitive is why we introduced reifiers and why I am very wary about triple terms (but see their effective use for those).
<pchampin> It *can* happen in Turtle, it is just unpleasant to write...
<niklasl> It's hard to skip that step while still introducing triple terms. But I do agree.
<pchampin> william_vw, that's a good summary
<pchampin> IMO
<pchampin> and also because it will come back and bite you (e.g. multiple mariages)
<pchampin> more pragmatically
<william_vw> thanks @enrico for the answer. sorry for being stunned.
So we have: rdf:reifies as a property and rdfs:Proposition as a class (which is the range of rdf:reifies); we do not have rdf:reifier or rdf:TripleTerm.
<niklasl> +1
<pchampin> +1
<niklasl> +1 to pchampin, triple terms denote resources of type rdfs:Proposition
<niklasl> tl see w3c/
<gb> Issue 61 Explain how classic RDF reification relates to triple terms and rdf:reifies (by niklasl)
Another proposal: triple terms denote instances of rdf:Proposition, and rdf:reifies has range rdf:Proposition
<william_vw> that makes a lot more sense to me
<pchampin> +1 to have this constraint in RDF entailment
triple terms are of type rdf:Proposition (in RDF semantics), and rdf:reifies has rdfs:range rdf:Proposition (in RDFS semantics)
<william_vw> +1
<pchampin> +1
We agree on: triple terms are of rdf:type rdf:Proposition (in RDF semantics), and rdf:reifies has rdfs:range rdf:Proposition (in RDFS semantics)
<tl> +1
<niklasl> +1
<enrico> +1
<gkellogg> +1
<pfps> +0.9
<pchampin> I have to jump now, sorry -- I'm +0 on introducing rdfs:state, (and -1 on making it the default for the annotation syntax, just in case, but IIUC this is not what is being discussed here)
<niklasl> For specific kinds of reifiers, you can use OWL to infer new triples, e.g. : <Alice> :bought <Lenny_the_Lion> from a Purchase reifier: w3c/
<gb> Issue 27 Integrating different ontology designs through entailment upon triple terms (by niklasl) [use case]
<william_vw> I have to leave - good weekend everyone
<gkellogg> <<( :s :p :o )>> a rdf:Proposition .