Explicit RDF property for "literal has datatype D"?

[ChristophLange]

Dear all,

  we are working on embedding RDF-based ontologies into another host language
(see below for background).  While saying that some RDF literal has a datatype
is hard-coded into the syntax of any RDF serialization, we would have to make
it explicit in our representation, and our representation of that would
actually be quite close to RDF triples.

Let me say it in terms of RDFS: There are classes rdfs:Literal and
rdfs:Datatype, and what I'm looking for is just a simple property

p a rdf:Property ;
  rdfs:domain rdfs:Literal ;
  rdfs:range rdfs:Datatype .

I could invent p myself, but maybe there is already some established ontology
that contains such a property p.  Do you know any?

Background, for those interested: We are working on improving the
documentation of ontologies.  As a means for that, we use the mathematical
markup language OMDoc, which supports literate-programming-like embedding of
documentation into formalized knowledge.  See
https://svn.omdoc.org/repos/omdoc/trunk/doc/blue/foaf and then
* eswc09-poster.pdf (poster abstract)
* eswc09-a0poster.pdf (poster)
* mkm09.pdf (full paper, but for a more mathematical audience)

Cheers, and thanks in advance,

Christoph

--
Christoph Lange, Jacobs Univ. Bremen, http://kwarc.info/clange, Skype duke4701

[Toby Inkster-4]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 6 Jul 2009, at 14:34, Christoph LANGE wrote:

> p a rdf:Property ;
>   rdfs:domain rdfs:Literal ;
>   rdfs:range rdfs:Datatype .


In terms of RDF, this can't be done. Well, the property could be  
defined using RDFS, but it could never legally be used. The  
definition would imply that "p" is used like so:

        "2008-01-01" p xsd:date .

However, RDF doesn't allow a literal to be used as the subject of a  
triple. Hence there is no property defined like the "p" you describe.  
There are supersets of RDF (such as Notation 3) which relax this  
restriction and would allow a property "p" like you describe to be used.

- --
Toby A Inkster
<mailto:mail@...>
<http://tobyinkster.co.uk>



-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.8 (Darwin)

iEYEARECAAYFAkpSEAQACgkQzr+BKGoqfTnHfQCfV50trF0sL9InwG08FgvQfbDQ
mBQAnA3dot8UHlT5bmvLtUycwQhAwYSB
=cB/I
-----END PGP SIGNATURE-----

[ChristophLange]

2009-07-06 17:24 Toby A Inkster <mail@...>:

> On 6 Jul 2009, at 15:53, Toby A Inkster wrote:
> > In terms of RDF, this can't be done. Well, the property could be
> > defined using RDFS, but it could never legally be used. The
> > definition would imply that "p" is used like so:
> >
> > "2008-01-01" p xsd:date .
>
> Forgot to mention... (somewhat perversely) the inverse is fine and
> dandy!
>
> ex:isDatatypeOf
> rdfs:domain rdfs:Datatype ;
> rdfs:range rdfs:Literal .
>
> xsd:date ex:isDatatypeOf "2008-01-01" , "2008-01-02" .

Indeed, Toby -- but then, one wouldn't be able to express the fact that
xsd:date is the datatype of the object of the triple

something dc:date "2008-01-01".

because the two literals "2008-01-01" merely have equal values, but there is
no notion of identity, as for resources.

Nevertheless, this restriction of RDF does not occur in my actual situation.
In the language that we are using to decorate ontologies, there is a facility
for annotating literals with further metadata (because the language has a
non-RDF origin; cf.  https://trac.kwarc.info/krextor/ticket/52). In _that_
place, I'd like to put the datatype information as a key/value pair, and now I
just need to come up with a good name for the key.

BTW, given that there is the class rdfs:Literal and the pseudo-attribe
rdf:value, which can be used creatively anyway
(http://www.w3.org/TR/2004/REC-rdf-primer-20040210/#rdfvalue), it would
actually be possible to model literals and datatypes as normal RDF triples:

y dc:date x .

x a rdfs:Literal ;
  rdf:value "2008-01-01";
  ex:datatype xsd:date .

Cheers,

Christoph

--
Christoph Lange, Jacobs Univ. Bremen, http://kwarc.info/clange, Skype duke4701

[Pat Hayes]

On Jul 6, 2009, at 9:53 AM, Toby A Inkster wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> On 6 Jul 2009, at 14:34, Christoph LANGE wrote:
>
>> p a rdf:Property ;
>>  rdfs:domain rdfs:Literal ;
>>  rdfs:range rdfs:Datatype .
>
>
> In terms of RDF, this can't be done. Well, the property could be  
> defined using RDFS, but it could never legally be used. The  
> definition would imply that "p" is used like so:
>
> "2008-01-01" p xsd:date .

rdfs:Literal is the class of literal VALUES, and bnodes and URIs can  
denote literal values. So you could use a bnode:

_:x p xsd:date .
_:x :seenAsLiteral  "2008-01-01" .

(I made up the second property, but you get the idea.)
OR, you could just ignore this damn silly syntactic restriction and  
put the literal in the subject, like many RDF systems now do.

Pat Hayes

>
> However, RDF doesn't allow a literal to be used as the subject of a  
> triple. Hence there is no property defined like the "p" you  
> describe. There are supersets of RDF (such as Notation 3) which  
> relax this restriction and would allow a property "p" like you  
> describe to be used.
>
> - --
> Toby A Inkster
> <mailto:mail@...>
> <http://tobyinkster.co.uk>
>
>
>
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.8 (Darwin)
>
> iEYEARECAAYFAkpSEAQACgkQzr+BKGoqfTnHfQCfV50trF0sL9InwG08FgvQfbDQ
> mBQAnA3dot8UHlT5bmvLtUycwQhAwYSB
> =cB/I
> -----END PGP SIGNATURE-----
>
>
>

------------------------------------------------------------
IHMC                                     [phone] or [phone]
40 South Alcaniz St.           [phone]   office
Pensacola                            [phone]   fax
FL 32502                              [phone]   mobile
phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes

[Jeremy Carroll-2]

Pat Hayes wrote:
>>
>>> p a rdf:Property ;
>>>  rdfs:domain rdfs:Literal ;
>>>  rdfs:range rdfs:Datatype .
>>
>
>
> _:x p xsd:date .
> _:x :seenAsLiteral  "2008-01-01" .
I tend to write these examples as

_:x p xsd:date .
_:x owl:sameAs  "2008-01-01" .

Semantically that has a literal as the subject, and it works around the
legacy syntactic restriction

Unfortunately the reasoning required to make this work means that simple
RDF systems may well not get it.

Jeremy
 

[Jeremy Carroll-2]

Christoph LANGE wrote:
>
> y dc:date x .
>
> x a rdfs:Literal ;
>   rdf:value "2008-01-01";
>   ex:datatype xsd:date .
>  
This type of design was considered by RDF Core in 2002/2003 and the
current design was preferred.
You can always drill down in the archives to find some of the discussion.

I think the current design involves less dependence on reasoning to get
things started.

Jeremy

[Graham Klyne-4]

Jeremy Carroll wrote:

> I tend to write these examples as
>
> _:x p xsd:date .
> _:x owl:sameAs  "2008-01-01" .
>
> Semantically that has a literal as the subject, and it works around the
> legacy syntactic restriction
>
> Unfortunately the reasoning required to make this work means that simple
> RDF systems may well not get it.

Hmmm... doesn't this lead to an inference something like xsd:date being the same
as xsd:string?

Well, not that exactly, but "2008-01-01" is a plain literal and hence
self-denoting.  Then the thing _:x denotes is hence (a) exactly "2008-01-01",
and (b) in the value space of xsd:date, which I'm not aware of including lexical
forms - is my birthday a string?

#g

[Steve Harris-11]

On 6 Jul 2009, at 18:43, Jeremy Carroll wrote:

> Pat Hayes wrote:
>>>
>>>> p a rdf:Property ;
>>>> rdfs:domain rdfs:Literal ;
>>>> rdfs:range rdfs:Datatype .
>>>
>>
>>
>> _:x p xsd:date .
>> _:x :seenAsLiteral  "2008-01-01" .
> I tend to write these examples as
>
> _:x p xsd:date .
> _:x owl:sameAs  "2008-01-01" .

did you mean:

_:x p xsd:date .
_:x owl:sameAs "2008-01-01"^^xsd:date .

otherwise you get some odd conclusions.

- Steve

--  
Steve Harris
Garlik Limited, 2 Sheen Road, Richmond, TW9 1AE, UK
[phone]  http://www.garlik.com/
Registered in England and Wales 535 7233 VAT # 849 0517 11
Registered office: Thames House, Portsmouth Road, Esher, Surrey, KT10  
9AD

[Jeremy Carroll-2]

Steve Harris wrote:

>
>>>>
>>>>> p a rdf:Property ;
>>>>> rdfs:domain rdfs:Literal ;
>>>>> rdfs:range rdfs:Datatype .
>> _:x p xsd:date .
>> _:x owl:sameAs  "2008-01-01" .
>
> did you mean:
>
> _:x p xsd:date .
> _:x owl:sameAs "2008-01-01"^^xsd:date .
>
> otherwise you get some odd conclusions.
>
> - Steve

I don't see any odd conclusions ... we have not said anything about the
realtionship between the subject and object of a p triple, other than
the usual they are related by the p-property.

Jeremy

[Pat Hayes]

On Jul 6, 2009, at 12:43 PM, Jeremy Carroll wrote:

> Pat Hayes wrote:
>>>
>>>> p a rdf:Property ;
>>>> rdfs:domain rdfs:Literal ;
>>>> rdfs:range rdfs:Datatype .
>>>
>>
>>
>> _:x p xsd:date .
>> _:x :seenAsLiteral  "2008-01-01" .
> I tend to write these examples as
>
> _:x p xsd:date .
> _:x owl:sameAs  "2008-01-01" .
>
> Semantically that has a literal as the subject, and it works around  
> the legacy syntactic restriction

True, and that is better.

>
> Unfortunately the reasoning required to make this work means that  
> simple RDF systems may well not get it.

Just doing sameAs substitution isnt itself a very heavy reasoning  
burden, though.

Pat


>
> Jeremy
>
>
>

------------------------------------------------------------
IHMC                                     [phone] or [phone]
40 South Alcaniz St.           [phone]   office
Pensacola                            [phone]   fax
FL 32502                              [phone]   mobile
phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes

[Pat Hayes]

On Jul 6, 2009, at 1:48 PM, Jeremy Carroll wrote:

> Steve Harris wrote:
>>
>>>>>
>>>>>> p a rdf:Property ;
>>>>>> rdfs:domain rdfs:Literal ;
>>>>>> rdfs:range rdfs:Datatype .
>>> _:x p xsd:date .
>>> _:x owl:sameAs  "2008-01-01" .
>>
>> did you mean:
>>
>> _:x p xsd:date .
>> _:x owl:sameAs "2008-01-01"^^xsd:date .
>>
>> otherwise you get some odd conclusions.
>>
>> - Steve
>
> I don't see any odd conclusions ... we have not said anything about  
> the realtionship between the subject and object of a p triple, other  
> than the usual they are related by the p-property.
>

Right. BTW, you could now say that the domain of p was  
rdf:PlainLiteral :-)

http://www.w3.org/TR/rdf-plain-literal/

Pat


> Jeremy
>
>
>
>

------------------------------------------------------------
IHMC                                     [phone] or [phone]
40 South Alcaniz St.           [phone]   office
Pensacola                            [phone]   fax
FL 32502                              [phone]   mobile
phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes

[ChristophLange]

2009-07-06 19:46 Jeremy Carroll <jeremy@...>:

> Christoph LANGE wrote:
> > y dc:date x .
> >
> > x a rdfs:Literal ;
> >   rdf:value "2008-01-01";
> >   ex:datatype xsd:date .
>
> This type of design was considered by RDF Core in 2002/2003 and the
> current design was preferred.
> You can always drill down in the archives to find some of the discussion.

Thanks for that pointer!  Indeed I found some interesting discussions on the
mailing list:

http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001Nov/0011.html
http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Dec/0007.html
http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Sep/0002.html
http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Jan/0472.html
http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001Nov/0295.html

As I found out, it was discussed for a while to use rdf:type in place of the
ex:datatype above, but then that was found to have undesirable consequences.

But now I know how _I_ will do it.  I will borrow rdf:datatype from RDF/XML
(http://www.w3.org/TR/rdf-syntax-grammar/#section-Syntax-datatyped-literals),
which is not a _property_ in the strict sense, but just fine for my purpose.
The purpose of my work is not to change the semantics of RDF, but to embed RDF
with its original semantics into a new XML-based host language in order to
have more expressive annotation facilities than in RDF/XML or XHTML+RDFa.  So
what I need for that is some _syntax_, and as a syntactic reference to the
datatype of a literal, rdf:datatype is perfect.

Cheers,

Christoph

--
Christoph Lange, Jacobs Univ. Bremen, http://kwarc.info/clange, Skype duke4701

[Alan Ruttenberg-2]

On Mon, Jul 6, 2009 at 4:54 PM, Pat Hayes<phayes@...> wrote:

>
> On Jul 6, 2009, at 1:48 PM, Jeremy Carroll wrote:
>
>> Steve Harris wrote:
>>>
>>>>>>
>>>>>>> p a rdf:Property ;
>>>>>>> rdfs:domain rdfs:Literal ;
>>>>>>> rdfs:range rdfs:Datatype .
>>>>
>>>> _:x p xsd:date .
>>>> _:x owl:sameAs  "2008-01-01" .
>>>
>>> did you mean:
>>>
>>> _:x p xsd:date .
>>> _:x owl:sameAs "2008-01-01"^^xsd:date .
>>>
>>> otherwise you get some odd conclusions.
>>>
>>> - Steve
>>
>> I don't see any odd conclusions ... we have not said anything about the
>> realtionship between the subject and object of a p triple, other than the
>> usual they are related by the p-property.
>>
>
> Right. BTW, you could now say that the domain of p was rdf:PlainLiteral :-)
>
> http://www.w3.org/TR/rdf-plain-literal/

Yes, except isn't that exactly wrong? At least by the desired sense
noted in the topic of the mail message "literal has datatype D".

A plain literal had datatype http://www.w3.org/TR/rdf-plain-literal/
which isn't any datatype other than itself. It certainly isn't and
xsd:date.

Is what's trying to be said that a certain string should be
interpreted as a certain datatype? But if that's the case, why not
just write it "certain string"^^datatype? (if it is malformed, you
will get an inconsistency, which is good!)

Sorry if I'm being dense.

-Alan

>
> Pat
>
>
>> Jeremy
>>
>>
>>
>>
>
> ------------------------------------------------------------
> IHMC                                     [phone] or [phone]
> 40 South Alcaniz St.           [phone]   office
> Pensacola                            [phone]   fax
> FL 32502                              [phone]   mobile
> phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes
>
>
>
>
>
>
>

[Jeremy Carroll-2]

Thanks Alan, I had made a mistake of not realising that the subject line
did specify stuff about the property p.

Alan Ruttenberg wrote:

>> On Jul 6, 2009, at 1:48 PM, Jeremy Carroll wrote:
>>
>>
>>    
>>> I don't see any odd conclusions ... we have not said anything about the
>>> realtionship between the subject and object of a p triple, other than the
>>> usual they are related by the p-property.
>>>
>>>
>>>      
> Yes, except isn't that exactly wrong? At least by the desired sense
> noted in the topic of the mail message "literal has datatype D".
>  

This suggests that we should have been talking all along about the
explicitly typed literal having datatype (or even just rdf:type) being D.

p a rdf:Property ;
rdfs:domain rdfs:Literal ;
rdfs:range rdfs:Datatype .
_:x p xsd:int .
_:x owl:sameAs  "2"^^xsd:int .

And possibly

p rdfs:subPropertyOf rdf:type .

Jeremy

[ChristophLange]

2009-07-06 23:47 Alan Ruttenberg <alanruttenberg@...>:

> > Right. BTW, you could now say that the domain of p was rdf:PlainLiteral
> > :-)
> >
> > http://www.w3.org/TR/rdf-plain-literal/
>
> Yes, except isn't that exactly wrong? At least by the desired sense
> noted in the topic of the mail message "literal has datatype D".
>
> A plain literal had datatype http://www.w3.org/TR/rdf-plain-literal/
> which isn't any datatype other than itself. It certainly isn't and
> xsd:date.

At least, rdf:PlainLiteral isn't what _I_ was searching for.  Alan, you got my
question right in the sense that I was asking for how to represent the
relation between a literal and its datatype for _any_ kind of datatype --
non-plain datatypes in particular.

Cheers,

Christoph

--
Christoph Lange, Jacobs Univ. Bremen, http://kwarc.info/clange, Skype duke4701

[Pat Hayes]

On Jul 6, 2009, at 4:47 PM, Alan Ruttenberg wrote:

> On Mon, Jul 6, 2009 at 4:54 PM, Pat Hayes<phayes@...> wrote:
>>
>> On Jul 6, 2009, at 1:48 PM, Jeremy Carroll wrote:
>>
>>> Steve Harris wrote:
>>>>
>>>>>>>
>>>>>>>> p a rdf:Property ;
>>>>>>>> rdfs:domain rdfs:Literal ;
>>>>>>>> rdfs:range rdfs:Datatype .
>>>>>
>>>>> _:x p xsd:date .
>>>>> _:x owl:sameAs  "2008-01-01" .
>>>>
>>>> did you mean:
>>>>
>>>> _:x p xsd:date .
>>>> _:x owl:sameAs "2008-01-01"^^xsd:date .
>>>>
>>>> otherwise you get some odd conclusions.
>>>>
>>>> - Steve
>>>
>>> I don't see any odd conclusions ... we have not said anything  
>>> about the
>>> realtionship between the subject and object of a p triple, other  
>>> than the
>>> usual they are related by the p-property.
>>>
>>
>> Right. BTW, you could now say that the domain of p was  
>> rdf:PlainLiteral :-)
>>
>> http://www.w3.org/TR/rdf-plain-literal/
>
> Yes, except isn't that exactly wrong? At least by the desired sense
> noted in the topic of the mail message "literal has datatype D".
>
> A plain literal had datatype http://www.w3.org/TR/rdf-plain-literal/
> which isn't any datatype other than itself. It certainly isn't and
> xsd:date.
>
> Is what's trying to be said that a certain string should be
> interpreted as a certain datatype? But if that's the case, why not
> just write it "certain string"^^datatype? (if it is malformed, you
> will get an inconsistency, which is good!)
>
> Sorry if I'm being dense.

No, but I was probably making a mistake trying to be humorous about  
rdf:PlainLIteral, sorry.

Would it be wrong? It depends on exactly what we are talking about  
here. If the subject of :p is intended to be the datatyped literal,  
then right, it would be wrong to say it was a plain literal. But if  
the subject is supposed to be the character string of the literal,  
then this might be one way to say that. Though even then, it would, I  
agree, be potentially misleading.

The best way, if you can't put the literal in subject position, would  
be what Steve suggested above.

Pat

>
> -Alan
>
>
>
>
>>
>> Pat
>>
>>
>>> Jeremy
>>>
>>>
>>>
>>>
>>
>> ------------------------------------------------------------
>> IHMC                                     [phone] or (650)494  
>> 3973
>> 40 South Alcaniz St.           [phone]   office
>> Pensacola                            [phone]   fax
>> FL 32502                              [phone]   mobile
>> phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes
>>
>>
>>
>>
>>
>>
>>
>
>
>

------------------------------------------------------------
IHMC                                     [phone] or [phone]
40 South Alcaniz St.           [phone]   office
Pensacola                            [phone]   fax
FL 32502                              [phone]   mobile
phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes