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

Re: LinkedData != RDF ?

From: Gregg Kellogg <gregg@kellogg-assoc.com>
Date: Fri, 20 May 2011 13:40:41 -0400
To: Dan Brickley <danbri@danbri.org>
CC: William Waites <ww@styx.org>, Linked JSON <public-linked-json@w3.org>
Message-ID: <582035EB-9768-49C8-B783-2AD65BE42A99@kellogg-assoc.com>
Well, my goal was to spark some discussion on what I see as the barriers to adoption of RDF, and at least I think I've accomplished that.

My thought was that, for serialization purposes at least, using a defined profile that creates a namespace for classes and properties within a specific lexical context is one of the good ideas from RDFa 1.1, and in a more limited sense, from the JSON-LD spec. Too many ontologies re-define common predicates (google rich snippets, Facebook Open graph, dc:title, ma:title, dc:created, doap:created, to name a few).

My assertion was that something like WebIDL could be used as an alternate way to define the semantics of classes and properties used within a profile. This could allow implementations of these semantics to be used within applications without having to rely on OWL reasoning. This does not mean that it is a replacement for RDFS/OWL, but provides an alternate way for developers to specify rules about data that could be implemented specific to the operating environment. It might even be possible to derive these interfaces from the ontologies used within a given profile.

RDF Projections [1] allow the construction of language-native objects based on the predicates actually used in data. My WebIDL suggestion essentially considers starting with a projection definition. If this could potentially be used to embody cardinality and type restrictions such as xsd:minInclusive/maxInclusive owl:allValuesFrom and so forth along with a certain amount of class entailment, the projections could be used to perform legitimate operations on data from within a given runtime environment.

To make the Semantic Web relevant to developers I think some outside-the-box thinking is in order. I didn't mean to imply that everything done in RDF should be thrown out, there's been far too much excellent work done for that to make sense. But RDF is just not catching on in a big way, and some discussion is appropriate to find ways to make it relevant.

(Funny, this started out as a way of talking about alternatives to RDF Lists for those who might want an alternative :)

> On May 18, 2011, at 9:23 PM, Manu Sporny wrote:
> 
> ...
> Maybe. There are some that really don't like RDF and see JSON-LD as a
> way of getting away from RDF. If those people don't care about RDF list
> semantics, they would adopt whatever list semantics were provided in
> JSON-LD.

Gregg

[1] http://www.w3.org/2010/02/rdfa/sources/rdf-api/#dfn-projection

On May 20, 2011, at 5:56 AM, Dan Brickley wrote:

> On 20 May 2011 14:31, William Waites <ww@styx.org> wrote:
>> * [2011-05-19 16:37:07 -0400] Gregg Kellogg <gregg@kellogg-assoc.com> écrit:
>> 
>> ] * The model should be based on the notion of graphs, similar to RDF, but
>> ] where the semantics are more Class/Object based, rather than predicate
>> ] based. (i.e., I define a class definition with specific properties and
>> ] class inheritance/implements more similar to Ruby/Python, rather then
>> ] being predicate based.
>> 
>> I'm not sure this is such a good idea.
> 
> Me neither. RDF takes control away from schemas, allowing multiple
> schemas to be mixed-in when describing some object, without need for
> permission, anticipation or coordination from the original schema
> writers. And when that results in too many namespaces at instance
> level, you can always create an indirection schema that defines
> subclasses/subproperties of common terms so that instance data can use
> a single flat namespace instead. Seems easier than throwing away the
> base level and starting from scratch...
> 
> Dan
Received on Friday, 20 May 2011 17:41:36 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 16:25:34 GMT