Re: [office-metadata] m:data-value / m:data-type only on anonymous RDFnode?

[Svante Schubert <Svante.Schubert@Sun.COM>]

John F. Madden, MD, PhD wrote:
> I agree this is confusing.
>
> It may be clarifying to consider the implementation in the RDFa 
> specification.
>
> In RDFa, they actually have two different versions of the one 
> attribute that we call "property".
>
> If the "property" has an object that is something other than a URI, 
> then RDFa uses the attribute "property".
> If the "property" has an object that is a URI, then RDFa uses the 
> attribute "rel".
In RDFa they have no other opportunity as to describe all RDF statements 
in content.
In the OpenDocument format I favor to keep as much metadata in the 
metadata files.

If I remember correctly was the original idea about 
m:data-type/m:data-value in the content that the RDF literal being 
expressed by m:about and m:property might need further rdf statements as 
it is hardly  referable from the RDF/XML file.
>
> So RDFa would represent the following two triples:
>
>     ex:aDocument dc:author "Svante Schubert"^^rdf:XMLLiteral       
> as      <element about="ex:aDocument" property="dc:author">Svante 
> Schubert</element>
>
>         whereas
>
>     ex:aDocument rdf:type    ex:ChangeProposal    as <element 
> about'"ex:aDocument" rel:"rdf:type" href="ex:ChangeProposal"/>
>
>
> When using "property", RDFa does give the option to override the 
> default datatyping of a string (by default as rdf:XMLLiteral) by use 
> of a "datatype" attribute:
>
>     ex:aDocument dc:date "20070101"^^xsd:Date       would be      
> <element about="ex:aDocument property="dc:date" 
> datatype="xsd:Date>20070101</element>
>
>
> So..........conclusion:
>
>  I think that we may be trying to do the same thing here---specify 
> some metadata about the OBJECT of the triple.
>
> However, since we only have the m:property element, we need to 
> consider three cases:
>
>     (1) the object is a URI
Is there a benefit, when not keeping these kind of statements in RDF/XML?
>     (2) the object is a typed literal
How is the object literal being typed? When the literal can never be the 
subject? (Aside of the following notation: 20070101"^^xsd:Date)
>     (3) the object is an "untyped" literal, i.e. an rdf:XMLLiteral (or 
> perhaps, if we like, an rdf:literal, or even an xsd:string --whatever 
> we define as our "default).
For me a string as default seems fine.
>
> Any parser that parses ODF in-content metadata is going to have to 
> decide which of these three cases obtains, in order to parse the 
> metadata properly.
>
>
> John
>
>
>
>
>
>
>
>     
> On May 9, 2007, at 9:57 AM, Svante Schubert wrote:
>
>> The question is what is the RDF statement m:data-value and 
>> m:data-type are part of?
>>
>> As a reminder the optional attributes m:data-value and m:data-type 
>> can occur on every element, where m:about/m:property can occur as 
>> well (therefore not on text:meta-field).
>>
>> <attribute name="m:data-type">
>>    <data type="anyURI"/>
>> </attribute>
>> <attribute name="m:data-value">
>>    <ref name="string"/>
>> </attribute>
>>
>> Each of them is an RDF predicate, the RDF object is anyURI in case of 
>> m:data-type and a string in case of m:data-value.
>> But what is the RDF subject, it can not be the literal (element 
>> content) itself, as a literal is never a RDF subject.
>>
>> I would think it is the element, which is anonymous in RDF (without 
>> an IRI).
>>
>> In this case, without using an xml:id to bind the OpenDocument 
>> element via the meta manifest to an IRI, the data-type and data-value 
>> will never be part of the larger RDF graph of the document.
>>
>> Is this correct?
>>
>> Svante
>>
>>
>