Re: {spam?} Re: [office-metadata] clarifying fields and metadata

[Bruce D'Arcus <bruce.darcus@OpenDocument.us>]

On Mar 13, 2007, at 10:48 AM, Svante Schubert wrote:

> Bruce D'Arcus wrote:
>>
>> Svante Schubert wrote:
>>
>>> Why not take 'meta:property' to describe the label you are using?
>>
>> Because it's semantics are different. It encodes a predicate, while  
>> here you want to say which predicate to use for display.
> Isn't this only a difference in RDF vocabulary?
> Does it makes any difference if meta:property describes the element  
> content to be a vCard:FamilyName or citation:FullQuote? Both times  
> metadata is being attached to the ODF element.

Tell you what: if Elias is fine with this, I won't protest. It's up to  
him as far as I'm concerned.

....

>> I'm not sure. Would you advocate *requiring* this?
>>
>> <text:meta-label  
>> meta:classification="http://opendocument.xml.org/meta/fields/ 
>> citation">
>> (
>> <text:meta:label
>>     meta:about="urn:isbn:98347823"
>>      
>> text:display-property="http://opendocument.xml.org/meta/fields/ 
>> citation#citation-full">Doe 1999</text:label>,
>> <text:meta:label
>>     meta:about="http://ex.net/1";
>>      
>> text:display-property="http://opendocument.xml.org/meta/fields/ 
>> citation#citation-short">1999</text:label>
>> )
>> </text:meta:label>
> Your example seems valid to me,

Right, but I'm saying a plug-in (or other code) might assume a default  
display property. So would you require it be explicit in the content?

> but might as well been written as (still using the discussed xml:id):
>
> <text:meta-label xml:id="myID"  
> meta:classification="http://opendocument.xml.org/meta/fields/ 
> citation">
> (Doe 1999, 1999)</text:meta:label>
>
> Providing similar RDF triple in RDF/XML without giving detailed  
> information about the subparts of the label as we won't specify the  
> subparts of the string "14th of December", neither.

Let me tell you my worry about this for my use case. I don't know if  
this worry is based on sound engineering principles, but it's a worry  
nevertheless.

To me, I really have the impulse to want the URIs that identify the  
source(s) and the optional parameters for the citation in content,  
since it's of critical importance to the logic of the field. In fact,  
the RDF/XML data can always be re-gathered if it gets lost somehow, so  
long as those URIs are there.

So I get really nervous with the idea of leaving the label in text as a  
completely dumb container.

To generalize my citation field (the one already approved) though,  
would mean a somewhat more complex field; something like:

text:meta-field
	text:field-source
	text:field-body

That way I could put multiple references to RDF subjects in the  
text:field-source element.

One of the reasons I floated this idea earlier is that it's also  
structurally similar one of the field structure in OOXML.

Now, no flames please: I just raise this issue because I want us to  
clarify the details. This gets into bigger principles, of course, like  
when the URIs and such should be in content, and when in RDF/XML.

Bruce