Re: [office-metadata] RDF/XML and XPath

[Michael Brauer <Michael.Brauer@Sun.COM>]

Hi Bruce,

Bruce D'Arcus wrote:
> 
> On Feb 9, 2007, at 12:28 PM, Michael Brauer - Sun Germany - ham02 - 
> Hamburg wrote:
> 
>>> In the preferred implementation, you don't care about the XML. You 
>>> load the RDF/XML into an in-memory model, and access triples as 
>>> needed. You come across a URI in a metadata field, for example, and 
>>> you simply access the triples for it.
>>
>> Maybe that's the misunderstanding. The text field that has the XPath 
>> expression is not a meta data field from the RDF perspective, so you 
>> don't come across it.
> 
> Ah; that indeed was a misunderstanding.
> 
> So then are you saying that in essence it is not in content metadata? 
> It's just a way to create and display XML data that happens to be RDF?

Yes. Or to be more precise. It's a way to display RDF data.
> 
> If that's the case, I don't see how it's our concern? I don't really see 
> what we would put in the ODF spec that would pertain to this, except 
> constraints on the syntax.

The use case is that you have for instance a vCard instance in your 
document, and you want to display some of its RDF objects (name, street, 
etc.) in you letter.
> 
> E.g. we've been asked to come up with a generic mechanism to describe 
> metadata and to associate it with content. Does XForms -- a UI 
> technology -- really fit our requirements? I know I've asked something 

Well, XForms has an UI part, that's true, but it also has a part that 
describes how data in a document is bound to (or associated with) an 
underlying XML model. That's the part we are using in ODF, but not the 
UI part.

> like this before, but it was when I was assuming something else.
> 
> Finally, WRT to the above, how would you then categorize a citation 
> field, which includes meta:resource references to RDF resources? Is that 
> in-content metadata? I'd say yes.

That's indeed an interesting question. My (maybe wrong) assumption would 
be that I have the bibliographic data of let's say a book in a RDF-XML 
stream in the document, and that a couple of its RDF object literals 
also occur in the content. I would not assume that these literals are 
resources, but again, I may be wrong.

The question is than: Do we copy the metadata triples in question into 
the content (and therefore duplicate the data), or do we only specify in 
the content that the data should be displayed?

My assumption would be that it is sufficient to display them. Displaying 
the data only may even be better than copying it, because this would 
allow corrections of the underlying bibliographic data, without having 
to update all other instances of the triple in question (only the string 
that is display has to be updated).

But we may also copy it (this would be meta data field, that I think we 
may want have additionally). But the question is then: How do we ensure 
consistency of the data, and what do we do with conflicts? One solution 
would be to ignore that issue (that means, it is up to an application 
that processes the meta data to decide what to do in this case, but not 
to the office applications). Or we add an explicit link to the original 
triple in the RDF-XML stream. And here I think using XForms bindings 
would be an option we should consider.

But to come back to your original question: If I understood you 
correctly, the triple you would like to add specifies its object by a 
resource. The subject would be another URI. For a text field, you need 
some text to be displayed. What would be this text?
> 
> But this then raises the question: when does one use XForms, and when 
> the metadata field? And why two approaches?

You use both together: XForms, if you want to display meta data only, 
and a meta data field, if you want to add a triple into the content.
And I would say to display the meta data is the better choice if your 
document contains a set of maybe even structured meta data, where you 
only want to display some of them. And to add a metadata field is the 
better choice if you want to add a single triple that has no relation to 
meta data in the RDF-XML streams.

I hope this helps.


> 
> Bruce
> 
Michael

BTW: I further believe that the XPath issue even is not as severe in 
practice as it may look like at the first glance. But that's something I 
would like to explain next week.