Proposal on new Meta Manifest due to last call's discussion

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

Elias,

on our last call, you proposed a change of the metadata manifest [1].
I stumbled into changes that add similar information, but might as well 
clarify our original metadata manifest (example attached as 
manifest_earlier.rdf).

As Bruce requested to rephrase the description of the meta manifest, I 
had to realize how none-verbose the Property Elements "mm:entry" and 
"mm:element" are.

For instance, the wording of the following RDF statement would be quite 
bumpy:
mm:Manifest mm:entry mm:Entry .

I tried to exchange the predicate to something existing and found 
http://dublincore.org/documents/dcmi-terms/#hasPart
mm:Manifest dcterm:hasPart mm:Entry .

This was much better, but the following still does not work:
mm:Manifest dcterms:hasPart odf:Element .

Therefore I did some further changes. Even if we name a file 
"manifest.rdf", we do not have to describe in RDF the Manifest, but may 
directly describe the ODF package the Manifest is about.
Therefore the above statement could be written equally as:
odf:Package dcterms:hasPart odf:File .

This would make a wording for the spec much easier - a good sign for the 
ease of understanding.
And now I would be able to give two new statements to solve our IRC call 
problem:

odf:File dcterms:hasPart odf:Element .
odf:File dcterms:references odf:Element .

As the odf:Element is only in content.xml and styles.xml I might say as 
well:
odf:Content dcterms:hasPart odf:Element .

Where odf:Content is subclass of odf:File.
Like
odf:Content a rdfs:Class; rdfs:subClassOf odf:File .
similar to
:Woman a rdfs:Class; rdfs:subClassOf :Person . [2]

Find a complete meta manifest example attached, which provides the 
information to get from a text:meta-field in the content to a RDF/XML 
file (Named Graph).

PS:
One nice side effect, we do not need the mm namespace any longer.
Furthermore, I exchanged the odf: namespace to 
xmlns:odf="http://docs.oasis-open.org/opendocument/meta/package#";
and suggest not to add a new namespace for m:, but reuse the existing 
meta: namespace for our four content attributes instead, see [3] for 
details.

Have a nice week-end,
Svante

[1] IRC log between Wed:17:39:33 to Wed:18:05:11
[2] http://www.w3.org/2000/10/swap/Primer
[3] 
http://lists.oasis-open.org/archives/office-metadata/200704/msg00032.html

manifest_earlier.rdf

manifest_updated.rdf