Re: [Sem] Syntax and model-theoretic semantics: a complete proposal

Antoine Zimmermann  wrote:
In section SEMANTICS:
- The way "RDF-star term" is defined, they include RDF-star triples, so the "or RDF-star triples" is redundent.

No, that’s a source of confusion also for some of your later comments; I probably should be more clear (but the syntax definition is correct, albeit terse).
An "RDF-star term” may be a ""quoted RDF-star triple term”, but not a “RDF-star triple". I have fixed the words in the grammar to be more clear.

- "in a concrete syntax" -> are there different semantics for different concrete syntaxes?

No, but concrete syntaxes may differ wrt having a label associated to bnodes — e.g., N-Triples forces you to always have a label, while Turtle does not.

- "RDF-star triples in a concrete syntax to their N-triples representation as string" -> N-triples, for the moment, does not have a representation for quoted triples.

I fully agree. I assume here that some syntactic transformation will be available at some point.

Also, it has multiple representations for some graphs in a concrete syntaxes, e.g., this Turtle string:

[] a [] .

has multiple N-triples representations.

I agree, and that is exactly the reason why L is not universally unique, but it is part of the interpretation: every interpretation may choose its own L mapping.

You may choose a normalised representation, but then, "[] a []" always denote the same thing everywhere, e.g.:

<<[] a []>>  :p  <<[] a []>> .

the subject and object denote the same thing.

See above: I don’t want to choose a normalised representation, since some  bnodes may not have a syntactic label to start with.

- [I+A] is ill-defined for RDF-star triples. If x = (s,p,o) is an RDF-star triple, it may be the case that [I+A](x) = TRUE or FALSE, or that [I+A](x) = IT([I+A](s),[I+A](p),[I+A](o)). You need a way to distinguish the function that maps terms to resources from the function that maps triples or graphs to {TRUE,FALSE}.

No. This stems from your confusion between quoted triple terms and triples.

- I find it strange that the "simple semantics" requires constraints on the interpretation of some specific IRIs (namely rdf:type and rdf-star:TEP)

That’s the choice of the CG final report, not mine :-)

I do not understand the utility of the examples. They are merely examples of the syntax of Turtle-star.

The utility is to show Turtle-star in this specific syntax.

2. the variant which distinguishes syntactic from semantic quoted
   triples (wiki: alt syntax specification),

Again: "RTD-star" -> RDF-star

thanx, fixed.

Again, the mapping L is not sufficiently well defined.
Again, the examples are just showing examples of a concrete syntax.

See above.

3. the variant with a quoting operator which gives a syntactic reading
   to arbitrary terms (wiki: alt syntax with quoted terms specification).

The mapping L is only used in the 5th point of [I+A] where r is necessarily a quoted term.

Indeed, that’s why it is used there...

L could be defined as a mapping from quoted RDF-star terms to strings. Or, even simpler, L could be eliminated and ITL be defined as a mapping from quoted RDF-star terms to IR. The intermediary mapping to strings does not affect anything.

Yes I could, but in order to formally define that mapping I would need to use L anyway, so…

cheers
—e.