– DRAFT –
RDF-star Semantics TF

i also wondered

can someone send the link again? I am searching

<AndyS> https://github.com/w3c/rdf-star-wg/wiki/RDF-star-%22liberal-baseline%22

<william_vw> en garde!

I simply like small models :)

<pfps> embedded triple terms don't count in the RDF Semantics?

we are just trying to define the semantics of the syntax we allow

<pfps> :x rdf:refies << :a :b << :c :d :e >> >> doesn't RDF-imply that << :c :d :e >> is an instance of rdf:Proposition

<Zakim> tl, you wanted to ask if we define what a reifier is, i.e. what it "means" that something rdf:reifies a triple term and to

<niklasl> And it doesn't imply that :d is a rdf:Property?

<niklasl> (nor :b)

<pfps> :john :believes << :a rdf:reifies :b >> doesn't make :b a rdf:Proposition

<tl> but we don't say that triple terms are true

<niklasl> Why not add a set IPR called the set of propositions of I; and a mapping from triple terms into IPR?

<Souri> Are we supporting nested triple-terms in RDF1.2 syntax?

<AndyS> souri - yes (as in "the current defn of the data model and the concrete syntaxes allow triple-terms inside triple-terms") https://www.w3.org/TR/rdf12-concepts/#section-triple-terms

<niklasl> @Souri Yes *at least* as nested objects (depending on if abstract syntax is isomorphic to N-triples, rather than isomorphic to symmetric RDF). That's a consequence of triple terms being triples (which can have triple terms). (I'm wary of even that but see theoretical benefits.)

I agree with pfps there, so why sould the range of predicate in a triple term matter

<enrico> :john :believes :b1.

<enrico> :b1 rdf:reifies <(:c1 rdf:reifies :d)>.

<enrico> --> :d rdf:type rdf:Proposition

<pfps> :john :believes <( :a rdf:reifies :b )> .

<enrico> --> :b rdf:type rdf:Proposition

:john :believes :b1.

:b1 rdf:reifies <(:c1 :a :d)>.

:a rdfs:range :G.

Do we get

:d a :G:?

<niklasl> We should not (as in do not want to); right?

<pfps> :John :believes <( :sally :loves : aaron )> .

<pfps> :loves rdfs:range :person .

<niklasl> That would imply that reasoning is over all "known" propositions rather than truths.

:b1 rdf:reifies <(:c1 :a :d)>.

makes: a a property? also if :a only occurs once in a nested position? (open question)

<Souri> IMO RDF1.2 syntax can be simplified if we use multi-triple way of expressing the same thing (using Doerthe's example from above): :john :believes :b1. :b1 rdf:reifies <<(:c1 :a :d)>> . INSTEAD OF => :john :believes <<( :b1 rdf:reifies <<(:c1 :a :d)>> )>> .

<niklasl> Believing in a proposition is like believing in a numbers. Only crazy people like me doubt their existence?

<pfps> It appears that Enrico wants :john :believes <( :sally :loves :aaron )> to simple-imply that :loves is in IP.

<AndyS> Is this the leakage? :s :p <<( :x :p :z )>> . :p rdfs:range :R => :z rdf:type :R .

<enrico> yes, andys

because it also leaks out

long ago

does not matter anymore

<tl> +1 to niklas

<niklasl> If it is true, as in asserted, as in "triple is in the graph", then entailment applies.

<Souri> +1 to enrico's comment about not having rdf:Proposition at all

<niklasl> <a-conflationist> :likes << rdf:type owl:sameAs owl:sameAs >> .

<niklasl> => the world is united in oneness

that was why I was so afraid of transparency, but then pfps version of "weak" transparency convinced me

<enrico> :p1 rdfs:subproperty :p2.

<enrico> :a rdf:reifies <(:b :p1 :c)>.

<enrico> does not entail :a rdf:reifies <(:b :p2 :c)>.

<niklasl> (Not known to be true.)

<tl> enrico i think this is really wrong, i.e completey unintuitive, i.e. it makes no sense

<william_vw> tl I believe the condition of the entailment rule simply does not apply

<william_vw> If <x,y> is in IEXT(I(rdfs:subPropertyOf)) then x and y are in IP and IEXT(x) is a subset of IEXT(y)

<Souri> Doesn't this mean that (b,c) in :p1 property in the context of :a? If so, wouldn't that imply (b,c) in :p2 in the same context? Or, we say that the context specific extension thing will not be considered.

<william_vw> (from https://www.w3.org/TR/rdf11-mt/)

<william_vw> but here, IEXT(x) is not being extended

niklas example (if I understood):

:loves rdfs:subProperty :knows.

:a rdf:reifies <(:bob :loves :sue)>; a :Lie.

should not imply

:a rdf:reifies <(:bob :knows :sue)>; a :Lie.

<niklasl> Yes; thank you doerthe!

<tl> @doerthe thanks! i think it *should* imply

Even here in the negated context?

<tl> i wonder what those all sorts of horrible things are that enrico thinks could happen

<william_vw> @tl it should not because the entailment does not apply

<william_vw> bob & sue are not part of the extension of loves

<william_vw> (they would be if asserted in the main graph)

<tl> oh, okay, now i get it - now it seems to me like i was wrong

<william_vw> @niklasl sorry for my relative radio silence on git - I had a paper deadline last night

<william_vw> I may not have time to give my 2 cents before Monday

<william_vw> I have to go everyone - happy holidays!

<niklasl> https://w3c.github.io/rdf-concepts/spec/

<niklasl> https://w3c.github.io/rdf-concepts/spec/#section-triple-terms-reification

I also need to go and wish you all happy holidays!

<Souri> Happy Holidays to everyone!