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

Re: TDL is compatible with tidy literals, even if current TDL MT is not (which should be fixed)

From: Patrick Stickler <patrick.stickler@nokia.com>
Date: Mon, 28 Jan 2002 12:25:02 +0200
To: Jeremy Carroll <jjc@hplb.hpl.hp.com>, Dan Connolly <connolly@w3.org>
CC: RDF Core <w3c-rdfcore-wg@w3.org>
Message-ID: <B87AF39E.C6DE%patrick.stickler@nokia.com>
On 2002-01-28 12:13, "ext Jeremy Carroll" <jjc@hplb.hpl.hp.com> wrote:

> 
> Hmmm ...
> 
> As I understand Patrick's position:
> 
> 
> There is an abstract layer over a tidy graph in which the TDL for one use of
> a literal node is distinct from the TDL for a second use.
> 
> This abstract layer seems to be a datatyping layer between the graph and the
> model theory.
> 
> I did consider this approach to doing the mathematics for TDL and couldn't
> get it to work, with all respect, Patrick has not shown to my satisfaction
> how it would work.
> 
> Let's suppose that it could be made to work.
> 
> I think an Occam's razor argument would favour a two layer approach over a
> three layer approach.

It would, *if* all things are otherwise equal. Occam's razor does
not say that simpler is always right if the simpler approach does
not provide a complete solution/explanation.

It appears to me that whether we have a two layer or three layer
approach depends directly on whether we have tidy or untidy
literals.

If literals remain untidy, then each literal node can be
interpreted as an actual lexical form (member of the lexical
space of a datatype) and that works. If, however, literals
are tidy, then literals can only be strings, and their
interpretation as a lexical form of some datatype must
happen somewhere at a higher level.

Finally, I'm not clear how the latter case constitutes
three levels, as if that interpretation is expressed in
MT, you just have the two levels, graph and MT. Right?
So how is it a level "between" the graph and MT?

Regards,

Patrick

 
> 
> 
> 
> 
> 
> 
>> -----Original Message-----
>> From: w3c-rdfcore-wg-request@w3.org
>> [mailto:w3c-rdfcore-wg-request@w3.org]On Behalf Of Patrick Stickler
>> Sent: 28 January 2002 09:07
>> To: Jeremy Carroll; Dan Connolly
>> Cc: RDF Core
>> Subject: TDL is compatible with tidy literals, even if current TDL MT is
>> not (which should be fixed)
>> 
>> 
>> On 2002-01-25 19:22, "ext Jeremy Carroll" <jjc@hplb.hpl.hp.com> wrote:
>> 
>>> 
>>> 
>>> 
>>>> 
>>>> 
>>>> Brace yourself for mind-bogglinly deep
>>>> formal argument:
>>>> 
>>>> premise:
>>>> <http://www.w3.org/> dc:title "W3C".
>>>> conclusion:
>>>> <http://www.w3.org/> dc:title "W3C".
>>>> 
>>> 
>>> 
>>> I am reviewing this.
>>> Currently I think Dan has a point.
>>> 
>>> So ...
>>> 
>>> So as is, this is a show-stopping bug on TDL.
>> 
>> Forgive me for slightly disagreeing with you Jeremy, but
>> the bug is not with TDL, but with the present MT definition
>> of TDL.
>> 
>> The basis of TDL is that one interpretes a given literal
>> as a lexical form of a datatype based on which TDL pairings
>> are defined or inferrable from the RDF graph.
>> 
>> In essence, TDL interpretation is like an axiom:
>> 
>> Per local idiom:
>> 
>> IF   a literal L is the object of the predicate rdf:value
>>      and the subject is an ononymous node with an rdf:type
>>      property defined,
>> THEN the datatype D is the object
>>      of that rdf:type property and the TDL is (L, D)
>> 
>> Per global idiom:
>> 
>> IF   a literal L is the object of a predicate other than
>>      rdf:value
>> THEN for each datatype D associated with the predicate
>>      via an rdfs:range property, there is a TDL (L, D)
>> 
>> and once we have a TDL, we know the value, based on
>> the definitions in the foundational DT MT and the
>> 1:1 relation between TDL pairings and mappings.
>> 
>> Note that in the above axioms, L is the actual literal
>> string, not the node, and D is the URI of the datatype
>> not the node/resource.
>> 
>> Neither of the above "axioms" require the graph to be
>> untidy for literals. The only thing that requires the
>> graph to be untidy is IFF the object node of a predicate
>> which either has a literal label or is an anonymous
>> node with an rdf:value defined literal is itself supposed
>> to denote the member of the value space. If the object
>> node need not denote the actual member of the value space,
>> then there's no problem.
>> 
>> See the attached illustration, which shows a literal tidy
>> graph where the literal "30" is interpreted as a lexical
>> form for two different datatypes. This shows that (a) the
>> TDL model itself is agnostic with regards to tidy versus
>> untidy literals, and also that the same literal (not
>> lexical form) may have multiple interpretations, depending
>> on the datatype associated with it.
>> 
>> Thus, I see no reason why the TDL MT cannot be revised
>> to allow tidy literals -- so long as folks are OK with
>> the axomatic like interpretation of TDL pairings and we
>> don't have to denote an actual value by any given literal
>> labeled node or rdf:value=literal qualified anonymous node.
>> 
>>> However it doesn't surprise me that there is a bug, very few of
>> us produce
>>> bug-free stuff first time.
>>> The normal response to bugs (even showstoppers) is to fix them,
>> so that's
>>> what I intend to do.
>> 
>> I have every confidence that the bug in the MT for TDL can
>> be fixed ;-)
>> 
>> Cheers,
>> 
>> Patrick
>> 
>> 
>> --
>> 
>> Patrick Stickler              Phone: +358 50 483 9453
>> Senior Research Scientist     Fax:   +358 7180 35409
>> Nokia Research Center         Email: patrick.stickler@nokia.com
>> 
>> 
>> 
> 
> 

--
               
Patrick Stickler              Phone: +358 50 483 9453
Senior Research Scientist     Fax:   +358 7180 35409
Nokia Research Center         Email: patrick.stickler@nokia.com
Received on Monday, 28 January 2002 05:24:00 EST

This archive was generated by hypermail pre-2.1.9 : Wednesday, 3 September 2003 09:44:01 EDT