Re: [office-metadata] Focus on model

[Michael Brauer - Sun Germany - ham02 - Hamburg <Michael.Brauer@Sun.COM>]

Hi,

(I'm sending this mail to the office-list, since I'm not a member of the metadata SC, but 
please don't hesitate to continue your discussion on the meta list.)

Lars Oppermann wrote:
> Florian, and all the others...
> 
> I did not follow the recent discussion in all its detail, please let me 
> know if I'm going into the wrong direction here.
> 
>  From what I have read during the last days, I was of the impression, 
> that marking fragments of OD content as subjects about which metadata 
> statements are made is a well agreed upon concept. I got the impression 
> that most of the problems were in cases, where OD content should be used 
> as the object of such a statement.
> 
> Consider the statement
> 
> {the_fragment-called_A; was_authored_by"; "John Smith"}
> {the_fragment-called_A; was_authored_by"; "John Smith"}
> 
> Assuming document content like this
> 
> <t:span xml:id="A">blah blah blah</t:span>
> 
> This could be encoded in an RDF/XML-like fashion as already noted like 
> this:
> 
> <rdf:RDF>
>   <rdf:Description rdf:about="#A">
>     <dc:author>John Smith</dc:author>
>   </rdf:Description>
> </rdf:RDF>

My understanding of this example is that the metadata shall be within another package 
stream than "content.xml". In this case, the relative IRI path for content.xml seems to be 
missing. If the metadata is contained in a stream next to content.xml, this would result in

<rdf:RDF>
   <rdf:Description rdf:about="context.xml#A">
     <dc:author>John Smith</dc:author>
   </rdf:Description>
</rdf:RDF>

The resolution of relative IRI paths within packages is already defined by the ODF 
specification. The only thing that is new is the fragment identifier that references an 
element within the stream, but this seems to be a common and well-understood XML technique.

I may be wrong, but I always thought that this is exactly the way how meta data is 
assigned to XML elements in general (it might be that we would have to use the rdf:ID 
attribute within the content.xml, but this shouldn't be an issue either).

It seems to me that another item that is discussed is whether the metadata should be 
within the content.xml or not. Well, since ODF already makes a separation between styles, 
content and metadata, I think it would follow the existing design principles to have it 
separate. But there is also a technicals reason why I think metadata at least optionally 
should be separate.  Metadata could be assigned to documents after they have been created. 
This in my opinion should be possible without altering the content.xml, provided that 
content.xml already contains IDs for those objects, that should get metadata assigned. 
Altering the content.xml for assigning metadata seems not only to be difficult, it may 
also break existing signatures.

I further believe that metadata support is easier to implement if metadata markup gets 
separated from the content markup, and if the only link between the two are IRIs 
(including fragment identifiers or xpointers).

Michael