W3C home > Mailing lists > Public > w3c-rdfcore-wg@w3.org > October 2002

Re: n-triples for datatype values [was: Agenda for RDFCore WG Telecon 2002-10-18]

From: Tim Berners-Lee <timbl@w3.org>
Date: Thu, 31 Oct 2002 19:14:45 +0100
Cc: "Brian McBride" <bwm@hplb.hpl.hp.com>, "Dan Connolly" <connolly@w3.org>, "Dave Beckett" <dave.beckett@bristol.ac.uk>, "Patrick Stickler" <patrick.stickler@nokia.com>, w3c-rdfcore-wg@w3.org
To: "Jos De_Roo" <jos.deroo.jd@belgium.agfa.com>
Message-Id: <A299CF48-ECFC-11D6-ABEB-000393914268@w3.org>


On Monday, October 21, 2002, at 08:38 PM, Jos De_Roo wrote:

>
> oops... forgot to say that we think that even if it
> would be described with extra triples e.g.
>   <#Jenny> <#age> [ xsd:integer "10" ] .
> or
>   _:x <#name> [ dt:string "chat"; xml:lang "fr" ] .
> an application is actually not forced to store the
> triples in between the [ and ] in case there is
> a (set of) unambiguous properties in there
> but instead could use them to construct a single
> *function* term or just store them as a [ whole ] term
> so why not use above as syntax for typed literals???
> (we have been using this trick over the past year
> or so, but actually we store both...)
>
> -- ,
> Jos De Roo, AGFA http://www.agfa.com/w3c/jdroo/
>
>
>
>                     Jos De_Roo
>                                          To:     "Patrick Stickler" 
> <patrick.stickler@nokia.com>
>                     2002-10-21           cc:     "Dan Connolly" 
> <connolly@w3.org>, "Dave Beckett"
>                     02:26 PM             <dave.beckett@bristol.ac.uk>, 
> "Tim Berners-Lee" <timbl@w3.org>,
>                                          w3c-rdfcore-wg@w3.org, "Brian 
> McBride" <bwm@hplb.hpl.hp.com>
>                                          Subject:     Re: n-triples 
> for datatype values [was: Agenda for RDFCore
>                                          WG   Telecon            
> 2002-10-18](Document link: Jos De_Roo)
>
>
>
>
>>> [...]
>>>
>>>>>>>
>>>>>>>   <a> <b> "foo"<dt1> .
>>>>>>>
>>>>>>> The latter is easier to parse.
>>>>>>
>>>>>> Oh! yeah verily.
>>>>>
>>>>> I was also wondering about that myself for similar reasons but
> didn't
>>>>> expect anyone else was worrying about such things, and wasn't going
>>>>> to propose it.
>>>>>
>>>>> +1 then
>>>>
>>>> I think that gives you enough of a mandate to do it that way in what
> you
>>>> write up.
>>>
>>> I understood DanC wanted something with a separator
>>> in between, e.g. <a> <b> "foo"^^<dt1>
>>> I then wonder where the possible langstring would
>>> fit, is it then "chat"-fr^^xsd:string or in N-Triples
>>> "chat"-fr^^<http://www.w3.org/2001/XMLSchema#string> ?
>>
>> This is still going to be a single node, right? Even when
>> N3 processes it?
>>
>> I'm not an N3 expert, but I'm presuming the above ^^ syntax
>> is not meant to force some kind of expansion into triples or
>> any other structure in terms of the abstract syntax, right?
>
> in that proposal it is a single node
> (which is not my own preference as you know;
> why else do we have RDF, interpretation props,
> etc.; having now up to 4 pieces of information
> in one node i.e. XML bit, lexical form, lang
> string and datatype-uri, not to mention the
> (un)allowable combinations is not very kiss)
>

To me, using "^^"   makes it clear that ^^ is a syntactic thing
whose semantics are in fact equivalent to "^"  except that
the formal triples representation is different.

So Jos, you can if you want dismantle the triple into two.
You will have a semantically equivalent graph.
Not the one which others want to be the canonical graph,
but one can define a Jos-canonical one.
They would entail each other.

I feel that "^^", being syntactic, should only be usable with a
fixed set of type URIs.

> -- ,
> Jos De Roo, AGFA http://www.agfa.com/w3c/jdroo/
>
>
Received on Thursday, 31 October 2002 13:16:00 EST

This archive was generated by hypermail pre-2.1.9 : Wednesday, 3 September 2003 09:52:36 EDT