Re: Summary: Annotation Syntax Proposals

Thanks, Andy! 

I wonder if the following are possible:

A)   :s :p :o  *{ :y :z } .
B)   :s :p :o  *{ :y :z }* .
C)   :s :p :o  {* :y :z *} .

Richard passingly proposed A a few weeks ago, together with the IMO rather convincing argument that it’s called RDF* after all.

Personally I like A. 

Given the very peculiar semantics of RDF* I’d strongly prefer if it didn’t use "@" but left that character for a semantics that treated attributed triples as belonging to the interpretation, not the syntax.

I tend to have difficulties reading {| … |} when the font size is small and generally I’m glad about every character I don’t have to type (or read).

Thomas



> On 19. Jan 2021, at 19:10, Andy Seaborne <andy@apache.org> wrote:
> 
> This is the action for issue 84.
> 
> Annotation syntax is additional syntax, for Turtle-star, to make it more convenient to write a triple and also add triples about that triple. It generates multiple triples.
> 
> A convenient way to write both:
> 
> :s :p :o .
> << :s :p :o >> :q .z .
> 
> It does not allow Turtle to say something that could not be said with the fundamental <<>> syntax.
> 
> The proposals are of the form:
> 
> :s :p :o annotationBlock? .
> 
> where the annotation block is some syntax and a predicateObjectList (Turtle grammar rule [7]) with <<>> as the subject.
> 
> 
> == Proposals
> 
> There have been several proposals, as well as more possibilities noted without advocacy.
> 
> 
> They fall into 2 categories: both have start-end delimiters and allow nesting, with different delimiters for start and for end.
> 
> 
> Type 1: Use { } with a introduction character.
> 
> Simple { } would work in Turtle because {} aren't used anywhere else but using the same syntax for SPARQL and N3 would be visually unclear.
> 
> So type 1 syntax uses an introduction character which as "@" - others have been suggested.
> 
> Example:
> 
> :subject :predicate :object @{ :source :URL } .
> 
> :subject :predicate :object @{ :source :URL ;
>                               :author "Alice"
>                             } .
> 
> 
> :subject :predicate :object @{ :source :URL ;
>                               :author "Alice" @{ :saidBy "Bob" }
>                             } .
> 
> "@{" is a new token in the grammar (no white space between the two characters.
> 
> 
> Type 2: Use distinctive enclosing delimiting pair, each is a new token.
> 
> One suggestion that works is "{|" ...  "|}"
> 
> "|" is one of several possible choices to go with "{", "}"
> 
> 
> :subject :predicate :object {| :source :URL |} .
> 
> :subject :predicate :object {| :source :URL ;
>                               :author "Alice"
>                            |} .
> 
> 
> :subject :predicate :object {| :source :URL ;
>                               :author "Alice" {| :saidBy "Bob" |}
>                            |} .
> 
> 
> == Requirements and notes
> 
> There are some requirements and constraints on the syntax design. Some are technically necessary and some are desirable. Visual appearance is also a factor.
> 
> 
> + SPARQL basic graph pattern matching uses a similar syntax to Turtle - it is "turtle with variables". Annotation syntax should fit naturally to being added to SPARQL so that it looks like Turtle.
> 
> SPARQL uses many extras characters not used for syntax in Turtle. These can be double up for annotation syntax (it appears to be technically possible) but it leaves the question of whether it is visually clear.
> 
> + Is nesting allowed in the annotationBlock? that is, be able to add annotations to the extra triples inside the annotation block? (Technical consequence: if not, then the Turtle grammar has to have more changes to exclude this possibility by having versions of some grammar rules with and without the annotation syntax, and also in the grammar rules a production uses. Annotation is part of predicateObjectList.
> 
> Nesting requires a delimiter style - mark the start and end of the annotation block.
> 
> + Other Turtle syntax
> 
> Turtle allows predicateObjectLists and objectLists.
> 
> A consequence of the annotationBlock is:
> 
> :s :p1 :o1 annotationBlock1? ;
>   :p2 :o2 annotationBlock2? .
> 
> :s :p :o1 annotationBlock1? , :o2 annotationBlock2? .
> 
> + RDF Collections (Turtle lists)
> 
> There has been no proposal for annotation of elements combined with list convenience syntax.
> 
> + Alignment to other languages
> 
> TriG and N3 are languages that share syntax features with Turtle.
> 
> Fitting into TriG is quite natural.
> 
> Of note in N3 is that "{ ... }" is already a language feature - a "formula".  It contains a number of N3-triples. It can go in the subject, predicate or object position.
> 
> N3 CG work: https://w3c.github.io/N3/spec/
> 
> Formula rule in the grammar:
> https://w3c.github.io/N3/spec/#grammar-production-formula
> 
> + Non-requirement
> 
> Annotation syntax is for Turtle-star, not N-Triples/N-Quads, in a similar way that the syntax "[ :p 123 ]" is not part of N-triples.
> 
> 
> 
>

Received on Tuesday, 19 January 2021 20:56:32 UTC