Re: @context file

Hi Tim,

> On 16 Aug 2015, at 24:18 , Timothy Cole <t-cole3@illinois.edu> wrote:
> 
> Ivan-
> 
> As discussed the changes to @context mean that agents creating JSON-LD can use type instead of @type and id instead of @id -- which is good, but what about when annotations stored as RDF are to be disseminated in JSON-LD?
> 
> I'm not clear from what I can discover from the JSON-LD Processing Algorithms and API document and from the test reports done for JSON-LD exactly how @context mappings are used when serializing RDF as JSON-LD (it does look that there is provision for applying @context, just not sure I understand all the rules, and when there is a choice -- as there would be for @type/type and @id/id, I would assume that the transforming agent has some discretion).

I do not really know. I *think* that this is entirely the discretion of the conversion tool; I would expect good tools to be able to take a @context file and make a maximum use of it. But I would actually be surprised if this was standardized. Well, maybe the framing tool do that, but those are not standard.

> 
> Is this another question for Greg, or do you or James know, or does someone else?

Asking Gregg is definitely a good idea. He knows JSON-LD inside out, having also make a complete implementation around it (and having a great experience in RDF tools, too).

> 
> Or maybe we have to provide libraries for this?
> 

I do not think so.

> Or maybe this is not an issue?

Well… I do not think this is an active issue for us.

@context is really there to simplify the JSON format of our data items. Where it *may* become an issue is if a fully RDF-based system has annotation data and wants to communicate/export the data with a pure JSON based annotations environment: they would have to export in the restricted JSON-LD format that we define. That may involve framing, etc, and that also means using @context. But that is mostly an implementation problem, not a specification one… (unless we want to define the details of framing in the standard, but I am not convinced we should do that). And I am not sure that scenario is a really realistic one, to be honest. Where I would expect LD to play a role is *consuming* existing annotation data into some LD environment (data integration with other types of data), where this problem does not occur, and not the other way round.

My 2 cents…

Cheers

Ivan

> 
> I ask not only as regards type and id, but because there is additional aliasing in @context we could consider to make the JSON-LD serialization seem more natural.
> 
> By the way, I appreciate you fixing up the Wiki page examples. Thank you.
> 
> -Tim Cole
> 
> 
> 
> -----Original Message-----
> From: Ivan Herman [mailto:ivan@w3.org]
> Sent: Wednesday, August 12, 2015 11:17 AM
> To: W3C Public Annotation List <public-annotation@w3.org>
> Subject: @context file
> 
> I have made the changes we agreed upon on the @context file, both on the github repo and on /ns.
> 
> Ivan
> 
> 
> ----
> Ivan Herman, W3C
> Digital Publishing Activity Lead
> Home: http://www.w3.org/People/Ivan/
> mobile: +31-641044153
> ORCID ID: http://orcid.org/0000-0003-0782-2704
> 
> 
> 
> 
> 
> 


----
Ivan Herman, W3C
Digital Publishing Activity Lead
Home: http://www.w3.org/People/Ivan/
mobile: +31-641044153
ORCID ID: http://orcid.org/0000-0003-0782-2704