W3C home > Mailing lists > Public > public-rdf-comments@w3.org > June 2013

JSON-LS Skolemization [was Re: JSON-LD Telecon Minutes for 2013-06-11]

From: David Booth <david@dbooth.org>
Date: Wed, 19 Jun 2013 14:01:39 -0400
Message-ID: <51C1F203.3070508@dbooth.org>
To: public-rdf-comments <public-rdf-comments@w3.org>, Andy Seaborne <andy@apache.org>
Hi Andy,

I just found out about this, which I had not previously seen because I 
cannot subscribe to the public-rdf-wg@w3.org list:
http://lists.w3.org/Archives/Public/public-rdf-wg/2013Jun/0112.html

 > From: Andy Seaborne <andy@apache.org>
 > Date: Sun, 16 Jun 2013 13:12:45 +0100
 > Message-ID: <51BDABBD.7090503@apache.org>
 > To: public-rdf-wg@w3.org
 >
 > On 11/06/13 17:43, Manu Sporny wrote:
 > > 7. Conversion of blank nodes to Skolemization IDs when
 > > going to RDF
 >
 > Please don't require this.
 >
 > Let the receiver of the (generalized) RDF decide how to
 > handle it.  Otherwise, systems that can handle partial or
 > full generalized RDF will not so easily be able to get such
 > generalised RDF.
 >
 > I hope to use a standard, high-quality JSON-LD library
 > and I don't want to have to use a modified version of it.
 > If the library is required-by-spec to skolemize, then it
 > hides the generalization.

You are saying that you want to use a JSON-LD library that generates 
invalid RDF?   I think that should be treated as a non-standard vendor 
extension.

Certainly non-standard extensions can be useful.  But it seems to me 
that the ability to generate "extended RDF" should be an *option* for 
the user to enable if desired.  The library should by produce valid RDF 
by default.

Would you agree?  If not, why not?

 > Skolemization by one component
 > is not easily reversible by another component.

It is if the skolem IRIs are generated to be RDF-standards-based round 
trippable.  That's the point of standardizing the use of a well-known 
URI suffix for skolemization

 >
 >
 > As far as I'm concerned, JSON-LD is RDF - a couple of
 > generalizations to bNodes here and there is not significant in
 > data model terms; inference can put bNodes anywhere anyway.
 > The major issue is breaking existing deployed software.
 > We can't assume the whole world upgrades to the new forms
 > when the specs are finalized.

That sounds like an argument in *favor* of requiring offending JSON-LD 
bnodes to be skolemized, so that existing deployed software that cannot 
handle, for example, bnodes in the predicate position, can make use of 
RDF that was serialized as JSON-LD.  Am I missing something?

David
Received on Wednesday, 19 June 2013 18:02:07 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:29:57 UTC