Re: JSON-LD requirements

On Jul 3, 2011, at 8:15 AM, glenn mcdonald wrote:

> Well, I object to all 11 of your Linked Data "requirements" in one way or another, before you even get to the JSON-LD derivations. I think they represent a very particular RDF-centric implementation of an idea that can and should be much simpler, and I'm with Kingsley in wanting to see the term "Linked Data" separated from the RDF mechanics.

Well, I was really attempting to do that. The term "RDF" doesn't appear anywhere in the document, and I was really trying to stick with (what I believe to be) a standard definition of "Linked Data" from Wikipedia and http://linkeddata.org. If we have a basic disagreement about what the term "Linked Data" means, then we can hardly agree what "Linked Data in JSON" should be.

If you have an alternative reference for a definition of "Linked Data" that is inconsistent with the one I've used, or if you feel that my assumed biases have influenced my attempt at an objective definition, please provide a reference and/or re-phrase the definitions to document.

>From a recent email from Kingsley [1]:

>> Linked Data is as I described above. de-referencable URIs for:
>> 
>> 1. Subject Name
>> 2. Attribute Name
>> 3. Optionally for Attribute Values.


I interpreted this to mean that the use of URIs as identifiers for SPO/EAV was implicit in LD. In fact, if anything, the requirements have left out the "de-referencable" part, where it would be expected that a URI object explicitly SHOULD be dereferenced to obtain further information (Linked Data Statements) about the referenced object.

(Kingsley, don't want to mis-represent you, please correct me if I didn't get this right).

> As we've said before, the question is what we're trying to accomplish here:
> 
> - serialize RDF in JSON for the benefit of people who already understand RDF but don't like the existing serializations for some reason
> 
> or
> 
> - encourage wider use of un-deconstructed graph representations of data by providing a simple serialization standard that takes advantage of JSON familiarity to reach a much larger audience of people who are motivated primarily by their own data and needs and don't necessarily yet even consciously think about graphs, much less understand RDF
> 
> I don't see the point of the former, so if that's what this "community" devolves into, I'll just unsubscribe. The latter, on the other hand, seems to me like a hugely valuable and important step that the *data* community should be trying to take.


I don't see the point of "Linked Data" in your alternative requirements. IMHO, the LD in JSON-LD is about having the represented information be _linked_ in the form of URIs, either to external information resources, or to descriptions contained within the source document. There may be value in having a JSON variant which represents unlinked graphs, but that might be for a different community.

Gregg

[1] http://lists.w3.org/Archives/Public/public-linked-json/2011Jun/0095.html

Received on Sunday, 3 July 2011 18:09:22 UTC