W3C home > Mailing lists > Public > public-linked-json@w3.org > February 2013

RE: Open Annotation / Default Context Location?

From: Markus Lanthaler <markus.lanthaler@gmx.net>
Date: Thu, 28 Feb 2013 12:36:26 +0100
To: "'Robert Sanderson'" <azaroth42@gmail.com>
Cc: <public-linked-json@w3.org>, "'Ivan Herman'" <ivan@w3.org>
Message-ID: <010501ce15a7$d88b1b80$89a15280$@lanthaler@gmx.net>
On Wednesday, February 27, 2013 9:54 PM, Robert Sanderson wrote:

> This is a great idea, which we're implementing at least until there's
> a better solution.
> The remaining question is whether it should be the recommended best
> practice?

I think it should.

> Ivan Herman said (quoted with permission), with regards to having
> somewhere on the W3C site to collect context documents from different
> ontologies:
> ----
> RDFa has the notion of initial contexts. They are much simpler and
> less powerful than @context, and they are essentially static files
> assigned to RDFa host languages, but they are more similar to @context
> files than vocabularies are. Those files are stored in
>     http://www.w3.org/2011/rdfa-context/
> Taking that example, we can set up, say,
>     http://www.w3.org/2013/json-context/
> or
>     http://www.w3.org/2013/json-ld-context/

Definitely prefer the second IRI for consistency reasons.

> as a more general resource where such files can be stored.
> ----
> This would fit more neatly with how we were expecting things to work,
> but certainly is not in the Open Annotation CGs remit to request :)
> Either this CG or the RDF WG would seem appropriate, if it's deemed to
> be the correct way forwards.

The situation is slightly different for RDFa because they mostly just
contain prefixes. I think what most users of JSON-LD want are terms. As long
as you use a single vocabulary that's no problem but as soon as you start
mixing them you will end up having collisions. Thus, instead of creating
profiles per vocabulary, I would expect that it's more useful to create
shared contexts for application domains. A JSON-LD document could them
signal compliancy to the conventions used for that domain by including a
specific IRI in the profile parameter. That's exactly what it's for.

Actually, it would make sense to define a JSON-LD context containing the
same IRI mappings as RDFa's initial context. This would allow you to easily
transform data from RDFa to JSON-LD and vice-versa.

Ivan, if you want to get this started, here's the RDFa JSON-LD context:

> Many thanks for your engagement with the issue, and the merged
> ontology/context suggestion :)

I'm glad I could help. In return, you could help by sharing our logo contest



Markus Lanthaler
Received on Thursday, 28 February 2013 11:37:00 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:53:21 UTC