W3C home > Mailing lists > Public > public-rdf-wg@w3.org > January 2013

Re: Using bnode identifiers for predicates, graph names

From: Sandro Hawke <sandro@w3.org>
Date: Thu, 31 Jan 2013 06:41:40 -0500
To: Pat Hayes <phayes@ihmc.us>,Andy Seaborne <andy.seaborne@epimorphics.com>
CC: public-rdf-wg@w3.org
Message-ID: <7288d8a1-2cf0-4bd8-8124-1ef48dd4c6d5@email.android.com>


Pat Hayes <phayes@ihmc.us> wrote:

>
>On Jan 30, 2013, at 2:24 PM, Andy Seaborne wrote:
>
>> 
>> 
>> On 29/01/13 17:09, Manu Sporny wrote:
>>> Hi folks,
>>> 
>>> I realize that this may open a can of worms, but we need a fairly
>>> definitive answer from the RDF WG on JSON-LD and using bnode
>>> identifiers for predicates and graph names.
>>> 
>>> One of the few remaining areas where JSON-LD currently deviates from
>>> RDF is by allowing, but frowning upon, the use of blank node
>>> identifiers as graph names and predicates. In both cases, blank node
>>> identifiers are scoped to the document.
>>> 
>>> It is our understanding that blank node identifiers weren't allowed
>>> for predicates because RDF/XML couldn't express that sort of markup.
>> 
>> And all the published versions of Turtle, N-triples etc do not
>support it.  A lot of existing code will depend on that.
>> 
>>> We were wondering if blank node identifiers for graphs had been
>>> discussed in detail.
>>> 
>>> We aren't interested in debating whether this is a good idea or not
>>> as that has a very high perma-thread potential. :)
>>> 
>>> We want to know if there is a known problem that cannot be worked
>>> around by allowing bnode identifiers to have document scope,
>allowing
>>> them to be used for predicates and graph names.
>>> 
>>> Thoughts?
>> 
>> Specifically on graph names -- document scoped bnodes identifers is a
>different issue and a good thing.
>> 
>> Yes - discussed at length in DAWG/SPARQL-1.0
>> 
>> The consensus reached was for IRIs only.  And in this WG has
>consensus on IRIs in datasets.  Not bnodes, not literals.
>> 
>> There is no practical experience of bnodes for graph names.  The
>semantic implications are "unclear".
>
>
>Any structure with bnodes is semantically equivalent to an isomorphic
>structure with different bnodes. This alone makes the idea of using
>bnodes as identifiers rather absurd. 
>


How is using bnodes to identify graphs any more absurd than using them to identify people (the canonical example)?    Blank nodes make prefect logical sense as local (file scope) identifiers.   They are clearly useful.

I'll agree with Andy's point, however, that this ship has already sailed.   While blank nodes are fine in any position in generalized rdf, they are not okay for predicates or graph names in standard rdf.

      - Sandro

>Pat
>
>>  I do not want to reopen that debate.
>> 
>> Known problem : does not work with other RDF systems.
>> 
>> 	Andy
>> 
>> 
>
>------------------------------------------------------------
>IHMC                                     (850)434 8903 or (650)494 3973
>  
>40 South Alcaniz St.           (850)202 4416   office
>Pensacola                            (850)202 4440   fax
>FL 32502                              (850)291 0667   mobile
>phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes

-- 
Sent from my Android phone with K-9 Mail. Please excuse my brevity.
Received on Thursday, 31 January 2013 11:42:55 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 16:25:53 GMT