- From: Patrick Stickler <patrick.stickler@nokia.com>
- Date: Fri, 11 Jan 2002 12:37:22 +0200
- To: ext Jan Grant <Jan.Grant@bristol.ac.uk>, Jeremy Carroll <jjc@hplb.hpl.hp.com>
- CC: Graham Klyne <GK@NineByNine.org>, Sergey Melnik <melnik@db.stanford.edu>, RDF Core <w3c-rdfcore-wg@w3.org>
On 2002-01-11 12:04, "ext Jan Grant" <Jan.Grant@bristol.ac.uk> wrote:
> On Thu, 10 Jan 2002, Jeremy Carroll wrote:
>
>> A few comments ...
>
> ...
>
>> The machinery I find hard to justify is:
>
> ...
>
>> - Always carrying the lexical values in the graph, and having the
> lexical
>> values in the model theory.
>
> Hear, hear. Lexical forms don't belong in a graph; values do. Getting
> from a serialisation to a graph is a parser issue (for some definition
> of "parser" that may include schema knowledge).
If you mean that members of the value space should have explicit
representation in the graph, then I disagree. To achieve representation
of actual values, you must have native support for all data types
for all statements, which is unreasonable and IMO unfeasible.
The RDF graph defines portable knowledge.
You may map that knowledge to other
representations specific to a given system, either in the API or
even some underlying implementation, but the graph represents
portable *system-independent* knowledge, not system specific
knowledge and therefore we have no recourse but to have lexical
forms in the graph, along with a consistent and unambiguous
interpretation of the values they represent.
There can be no native data types in RDF, only a consistent means
for declaring the data type of a lexical form.
Patrick
--
Patrick Stickler Phone: +358 50 483 9453
Senior Research Scientist Fax: +358 7180 35409
Nokia Research Center Email: patrick.stickler@nokia.com
Received on Friday, 11 January 2002 05:36:46 UTC