– DRAFT –
(MEETING TITLE)

<niklasl> A thought: https://gist.github.com/niklasl/3e8f972b9c1b5974982836d94b818474

<doerthe> what about naming it NewClass as a class for now and postpone the name discussion?

<doerthe> Just as a note: we need to change rdfs14 as well

<Zakim> thomas, you wanted to say that we need to talk about triple terms in subject position

<AndyS> https://www.w3.org/TR/rdf11-concepts/#dfn-rdf-statement

<niklasl> Proposition is minutely more precise in the "logic community", I think, just because of https://en.wikipedia.org/wiki/Proposition I think we only *need* to not have semantic drift of "proposition" in the docs. Statement is used mostly as "proposition", but sometimes as "the act of assertion", and sometimes "when made true" (to wobbly paraphrase

<niklasl> it).

<pfps> Since at least 1.1, an RDF statement is, in essence, the assertion of an RDF triple.

<pfps> "Asserting an RDF triple says that some relationship, indicated by the predicate, holds between the resources denoted by the subject and object. This statement corresponding to an RDF triple is known as an RDF statement."

<niklasl> https://gist.github.com/niklasl/3e8f972b9c1b5974982836d94b818474

<pfps> What needs to be changed, I think, is "A triple term is not necessarily asserted, allowing statements to be made about other statements that may not be asserted within an RDF graph."

<pfps> This uses "statement" as something that is not an RDF statement.

<Zakim> pchampin, you wanted to question whether RDF 1.1 used "statement" consistently

<pfps> So the problem is with rdf:Statement?

<niklasl> It's certainly made worse because of that.

<Zakim> doerthe, you wanted to mention that I see that proposition (or whatever we name it) is very similar to property

<thomas> type occurrence

<thomas> asserted triple reifier

<thomas> not asserted rdfs:Proposition rdf:Statement

<thomas> triple term

<niklasl> "Statement in the graph [which encodes the world]" is an obvious "map as the territory" fallacy though? An rdf:Statement may be in the graph, as the abstract encoding *in* the world. That is a reifier (specifically a token, adequately typed rdf:Statement).

<thomas> oups, what i just posted is wrong, a reifier is not asserted

<pchampin> enrico no, because the instances of rdf:Statement are much more like reifiers than like triple terms

<niklasl> "stated propositions"

<niklasl> (encoded by "triples in graphs")

<niklasl> interpretation ::= fact* # "stated proposition", that which "holds in the world"

<niklasl> fact ::= proposition

<niklasl> proposition ::= resource property resource

<niklasl> resource ::= literal value | proposition | other entity

axiomatic rriple rdf:statement rdfs:subclass rdfs:proposition.

<thomas> next try for a matrix:

<thomas> type occurrence

<thomas> asserted statement … rdf:type rdfx:Stated

<thomas> not asserted triple reifier (almost the same as rdf:Statement)

<thomas> Asserting an RDF triple says that some relationship, indicated by the predicate, holds between the resources denoted by the subject and object. This statement corresponding to an RDF triple is known as an RDF statement.

<thomas> from RDF Concepts 1.1 https://www.w3.org/TR/rdf11-concepts/#resources-and-statements

5.3.1 rdf:Statement

rdf: Statement is an instance of rdfs:Class. It is intended to represent the class of RDF statements. An RDF statement is the statement made by a token of an RDF triple. The subject of an RDF statement is the instance of rdfs:Resource identified by the subject of the triple. The predicate of an RDF statement is the instance of rdf:Property

identified by the predicate of the triple. The object of an RDF statement is the instance of rdfs:Resource identified by the object of the triple. rdf:Statement is in the domain of the properties rdf:predicate, rdf:subject and rdf:object. Different individual rdf:Statement instances may have the same values for their rdf:predicate, rdf:subject and

rdf: object properties.

<doerthe> (reading up right now, therefore no reaction yet))

<niklasl> "asserting a triple" implies that not all triples are asserted?

<niklasl> Not "is a triple", "denoted by a triple [term]".

<doerthe> OK, enrico, I have to think, but I think I am fine with your idea... when we discussed, I had not re-read the current definitions of rdf:Statement

Option 1: keep rdfs:Proposition, and add the axiomatic triple: rdf:Statement rdfs:subclass rdfs:Proposition.

option 2: replace rdfs:proposition with rdf:Statement.

I'd be in favour of option 2, just to be quick.

<niklasl> -1 for option 2, because it conflates the two

<AndyS> I lean towards option 1. Legacy rdf:Statement, use "statement" (natural language)

<doerthe> that is an advantage :D

<niklasl> I tend to think so to (an advantage) since it is the meaning itself, not its *role* as constituting truth ("the world").

<pfps> I like option 1-, i.e., no axiomatic triple

<Souri> I'd argue for ==> Triple = RDF Statement = Asserted Propositions. Also, rdf:Reifier owl:equivalentClass rdf:Statement . "Triple Structure" refers to a way of expressing an RDF proposition.

<niklasl> I'm also not sure about the axiomatic triple (yet), so rather option 1- as @pfps said.

<pfps> A problem with using rdf:subject is that its domain is rdf:Statement

<niklasl> See also w3c/rdf-semantics#61

<gb> Issue 61 Explain how classic RDF reification relates to triple terms and rdf:reifies (by niklasl) [spec:editorial]

<doerthe> that was my quaetion, pfps

<doerthe> question

<doerthe> is it a problem? (I think it is but did not think it through)

axiomatic triple: rdf:Statement rdfs:subClassOf rdfs:Proposition .

<thomas> +1 to that (although i need to see a complete proposal, covering all aspects, to be sure)

<niklasl> I am *very* cautiously positive to that. But it's a "would like to", not sure if it actually fully clarifies.

<Zakim> enrico, you wanted to argue why rdf:reifies is weaker within triple terms

<niklasl> To be clear for the minutes, we are settled on rdfs:Proposition (RDF Schema namespace), so w3c/rdf-schema#31 needs to be updated once it's reopened (it says it's temporarily closed).

<gb> CLOSED Pull Request 31 add rdf:dirLangString, rdf:reifies, rdf:Proposition, and modify rdf:JSON (by domel) [spec:substantive]

<Zakim> Souri, you wanted to ask ==> Since we already have rdf:Property, why not use rdf:Proposition (instead of rdfs:Proposition)?

<doerthe> ah, that would be a fix :)

<niklasl> I'm fine with it mostly because of rdfs:Resource and rdfs:Literal

<thomas> +1 to that

<doerthe> I like the idea to make it total

<pfps> w3c/rdf-semantics#69

<gb> Issue 69 modification to new semantics to permit entailment that every triple term is a proposition (by pfps)

<pfps> 1/ Triple terms in subject position is not very important in Semantics.

<pfps> The difference to Semantics is very much like the situation with respect to literals in subject position.

<pfps> The way forward, I think, is for the WG to decide which way to go and then Semantics can be easily fixed up.

<pfps> 2/ Semantics uses rdfs:Proposition as the type of triple terms.

<pfps> The name is arbitrary and could be changed to any "fresh" name with no other consequences.

<pfps> Some wording in Concepts concerning "statement" should be modified.

<pfps> The confounding fact is that rdf:Statement already exists and can't be changed.

<pfps> 3/ RDF Semantics has RE being a partial mapping from IR x IP x IR into IR.

<pfps> This causes problems with entailment and needs to be fixed.

<pfps> 4/ Entailment rules need to be adjusted.

(3) u=is fixed in the new version

<AndyS> 4/ Entailment rules in RDF Semantics need to be adjusted to align with the baseline rules.

<AndyS> 2 is resolved (today) and can be removed.

<AndyS> DRAFT: Accept: https://github.com/w3c/rdf-star-wg/wiki/RDF-star-%22liberal-baseline%22/eaa9863a2a21623fd94a34fc8e7484e596476460 with the list above (1,3,4) added as notes.

<niklasl> I have a hard time following the total RE argument. Does it also mean that: If { _:r1 rdf:reifies <<(:a :b :c)>> . :a owl:sameAs :A . } + (some entailment regime) => { _:r1 rdf:reifies <<(:A :b :c)>> . } mean that <<(:A :b :c)>> a rdfs:Proposition must be true in the LHS...?

<niklasl> The domain of RE cannot be "papered over", and may lead to broken RDF or RDFS semantics?

<pfps> If RE is partial then you lose observable consequences.

<pfps> If RE is total on IR x IP x IR I'm not sure what happens.

<pfps> You do lose some stuff in the simple semantics, but it might all be non-observable.

<niklasl> w3c/rdf-semantics#69

<gb> Issue 69 modification to new semantics to permit entailment that every triple term is a proposition (by pfps)

<niklasl> Perhaps solvable by A) dropping rdfs:Proposition, B) IP as subset of IR, maybe C) rdf(s):Proposition not a subclass of rdfs:Resource.

<AndyS> Blocking: w3c/rdf-semantics#69

<niklasl> IP must be equal to IR even for simple semantics for e.g. <<(:a rdfs:subPropertyOf :b)>> ?