Re: JSON-LD Basic

On 6/28/11 3:26 AM, Manu Sporny wrote:
> On 06/24/2011 02:48 AM, Kingsley Idehen wrote:
>> Some fundamental points that don't come through via the abstract:
>>
>> 1. Syntactic incompatibility != conceptual incompatibility
>> 2. Microdata, Microformats, RDFa all provide syntax for embedding graphs
>> within HTML documents
> I'll try to make these two more clear in the spec.
>
>> 3. Linked Data is about using de-referencable URIs in object (resource)
>> description (respresentation) oriented graphs where each URI resolves to
>> a representation of its referent i.e., graph based structured data
>> representation via hyperlinks.
> I don't think beginners are going to understand a sentence like this.

Of course not, I just sent you the gist of what needs to be written in 
more structured form with spec audience in mind.

>   I
> think I know what you're getting at though and will try to figure out
> some language to make it a bit more clear to a beginner.
>
> Keep in mind that while you're focusing on the prose, I didn't when
> drafting JSON-LD Basic. I just paid attention to what Glenn, Brian
> Peterson and you were saying and tried to accomplish as many of those
> goals as possible (in the limited amount of time I had available that day).

Yes, but you have to be able to make a spec based on a clear goal. The 
trouble is the goal (at least to me) isn't clear. Success really means 
decoupling JSON-LD from RDF. Otherwise, the spec should be called: 
JSON-RDF. Trouble, when you have JSON-RDF how would you explain its 
purpose in life bearing in mind existing RDF/JSON efforts?

You got to take RDF out, to be blunt.

>> If the critical points above are easily discerned from the abstract, we
>> finally have what I think most people seek: a simply spec for making
>> Linked Data via a JSON based syntax.
> My read is that you'd like it to talk about RDF as little as possible.

As per comment above, not at all. As I've stated repeatedly, any RDFer 
worth his or her salt can make RDF (with varying degrees for semantic 
fidelity) graphs from JSON-LD and an array of other data formats.
> Thinking out loud now:
>
> I think we could not mention RDF at all in the JSON-LD Basic document...
> perhaps there are two document conformance levels (Basic and Advanced)
> and one processor conformance level (Advanced)?

Too complex.

The spec should be about how you make Linked Data Graph (WWW age Linked 
Data Structures) using EAV/SPO triples via JSON. What's even more 
important than RDF is getting the narrative straight from the onset re. 
the use of URI based Names that resolve to Representations of their 
Referents.
> I think that we may be able to pull that off without causing any
> upwards/downwards compatibility problems.

An EAV/SPO based graph is inherently upward compatible and universally 
palatable :-)

> Basically, with JSON-LD Basic, there is one primary kernel of truth:
>
> If you have a valid @context, you have Linked Data.

No, that's a political slight of hand dressed up in syntax.

You have Linked Data because you constructed a Linked Data Resource 
using JSON. The EAV/SPO graph includes the use of URIs that Resolve to 
Representation of their Referent. The Representation of the Referents 
(esp. if Names are HTTP URI based) are negotiable.

The representation of a Data Object takes the form of one or more 
EAV/SPO patterns contained in a document (the Resource) available at an 
Address. Thus, you can refer to stuff by name and de-reference its 
medium specific representation, via an address. The name and the address 
are distinct, but this distinction just doesn't need to be confusing.
> What your Linked Data means is fully up to the @context document... and
> most people don't have to care about that at all.

Confusing to say: what your Linked Data means ...

The Linked Data has to exist. JSON-LD should be about creating Linked 
Data using JSON.

>   They just take their
> JSON markup, add a context and they're done. They know that anybody else
> that uses that same context with the same attributes expresses the same
> information. Full stop.

But they won't actually get to the place you describe above. That's the 
problem.

> The JSON-LD Advanced spec outlines how to take this JSON and convert it
> to RDF (for those that care about that sort of thing).

Why make a spec for RDF mapping when a simple Graph based on EAV/SPO is 
basically RDF modulo fidelity for handling blank nodes, data types, and 
locale matters? It will only be perceived as an RDF rehash in JSON. Then 
to make matters worse, you'll have to answer the question: what's wrong 
with the existing RDF/JSON efforts? Are they not vehicles for Linked 
Data too? Basically, just more unnecessary confusion and a strange 
degree of repetitive infighting. Do remember, RDF/JSON has existed for a 
while and I am yet to hear about what its flaws are re. Linked Data. 
Basically, how do you prove that relative to RDF/JSON this whole JSON-LD 
affair isn't good old NIH?

> The JSON-LD
> Advanced spec also deals with some of the more advanced technical
> topics, like normalization and digital signatures on graphs, that
> PaySwarm needs to succeed.

Kinda irrelevant to the spec. It shouldn't be about how any of our 
products will succeed.
> Now, we may need to change some JSON-LD Advanced bits and pieces to
> ensure that the above holds true, but does that seem like a better
> direction to you?

JSON-LD advanced by definition is contradictory if you are seeking mass 
comprehension of a lightweight spec for Linked Data construction via 
JSON :-)


Kingsley
> -- manu
>


-- 

Regards,

Kingsley Idehen 
President&  CEO
OpenLink Software
Web: http://www.openlinksw.com
Weblog: http://www.openlinksw.com/blog/~kidehen
Twitter/Identi.ca: kidehen

Received on Tuesday, 28 June 2011 08:52:00 UTC