Re: JSON-LD Syntax request for FPWD via RDF WG

Hi Manu,

On 24 May 2012, at 02:14, Manu Sporny wrote:
> Could you answer these questions, it would help me understand where
> you're coming from:
> 
> 1. What is the goal with making a "formal" connection to RDF?

The goal is that someone who looks at the spec, wondering how it can be used to serialize RDF into JSON-LD, or to obtain RDF from a JSON-LD document, will be able to find an answer, and a normative one.

In the version that you submitted to the WG, someone looking for this will come up empty:

1. Neither the abstract not the introduction mention RDF.

2. RDFa and Turtle are mentioned in the ToC, but all one can find there is an appendix “Markup Examples” and it has some subsections with examples that shows how JSON-LD can express similar stuff to Turtle and RDFa.

3. From the examples it is unclear if they are supposed to merely illustrate that JSON-LD has similar capabilities to Turtle or RDFa, or to illustrate that there are actually detailed algorithms for the conversion. These subsections don't link to the JSON-LD API.

4. So from reading all the RDF-related material in the JSON-LD syntax spec, one doesn't get any indication that there is another spec somewhere that normative defines the detailed algorithms.

> 2. What would a "formal" connection to RDF look like?

Have a normative section or subsection or appendix called “Conversion to and from RDF” that either has the detailed to- and from-RDF algorithms, or states where they are found with a normative reference.

This section can contain all sorts of disclaimers that you don't need RDF to use JSON-LD if that helps make it more palatable.

To further improve the usability of the spec for RDF-affine audiences, it should be mentioned somewhere in Section 1 or 2 that a detailed mapping to RDF exists. This can be one sentence with a link.

The main point being that someone who eyeballs the Intro or the ToC for mentions of RDF should be able to easily find some normative text that defines, or links to, the detailed to/from algorithms.

> 3. Where does it state that a "formal" connection to RDF is a
>   requirement for publishing a document via RDF WG? Why isn't a
>   number of examples and language making it clear that JSON-LD
>   can represent RDF enough?

Because RDF-WG has a charter that describes the scope of the WG. Having a formal connection to RDF and speaking to a particular item listed in the charter makes crystal clear why a certain piece of work is in scope. Having no formal connection makes it very easy to argue that a piece of work is out of scope. The claim that JSON-LD can represent all of RDF needs to be verifiable—I don't think that stating the claim and giving some examples is enough.

Best,
Richard

Received on Thursday, 24 May 2012 14:25:33 UTC