W3C home > Mailing lists > Public > public-linked-json@w3.org > October 2011

Re: JSON-LD Java Implementation feedback.

From: Stephane Fellah <fellahst@gmail.com>
Date: Mon, 24 Oct 2011 12:07:54 -0400
Message-ID: <CALfZuNpGiEgUwqc=sw1u78XxaKgKDoxVF3yfVVuVa47Xy3S+2Q@mail.gmail.com>
To: Markus Lanthaler <markus.lanthaler@gmx.net>
Cc: Linked JSON <public-linked-json@w3.org>, Fabian Christ <christ.fabian@googlemail.com>

My comments are inline.

On Mon, Oct 24, 2011 at 4:50 AM, Markus Lanthaler

> > Another point is that I always thought that JSON-LD should be easy to
> > interpret for people who are not that familiar with semantic web
> > technologies. It should be just JSON with extras. For those people, it
> > should be really easy to forget the @context and just grab the needed
> > values from the properties. But now there are these complex rules for
> > IRIs with and without prefixes that adds complexity. And you can only
> > understand this when reading the spec really carefully. One step to
> > make parsing easier would be the solution with @prefix and @token. It
> > just makes a complicated rule easy to understand.
> It's difficult to say what will be easier for people :-) But you are right.
> CURIEs (which we know just call prefixes for simplicity) where one of the
> most controversial topics. That's also the reason why we classified them as
> an Advanced Concept.
> I personally think that prefixes won't be used that often in JSON-LD
> anyway. Most people will stick to simple terms for JSON keys as they do now.
> I have yet to come across a JSON document using namespaces.
> To export RDF to JSON-LD, most of the models I export in my application
have already a predefined prefix mapping. It is easier for me to map the
json keys to curie form (using prefix) as it does not require for me to
define tokens every time I have a name clashing between different
ontologies. For an export scenario, use of qname as json keys would be the
majority of the case. However if you want to work from existing JSON format
to RDF, the use of token mapping to IRI would the majority of the case. I
think both will be used. I still think that we need to distinguish both
prefix mapping and token in the context, as it simplifies the
implementations of JSON-LD processor and also allows to the preserve the
original prefix mapping defined along the RDF document.

Stephane Fellah

> --
> Markus Lanthaler
> @markuslanthaler
Received on Monday, 24 October 2011 16:08:25 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:53:18 UTC