Re: [office-metadata] Metadata only for Content

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

Hi Bruce, Hi group!

Bruce D'Arcus wrote:
>
> On Nov 1, 2006, at 2:40 PM, Svante Schubert wrote:
>
>>     3     No XML elements describing Styles or Metadata will be part 
>> of this subset
>
> I think this is fine for now, though we may need to revisit.
Please do not hesitate to state your concerns right now. Designs are 
always closely developed to their requirements. Therefore the more we 
specify now, the sooner we are able to adapt, the less we have to spend 
time in the future.

>
> Metadata about metadata is the named graph approach that I know Dan 
> Brickley and John Barstow were both thinking about. For example, let's 
> say you register a metadata graph in the manifest like so:
>
>  <manifest:file-entry
>     manifest:media-type="application/rdf+xml"
>     dc:source="http://ex.net";
>     manifest:full-path="meta.xml"/>
>
> The dc:source attribute is metadata about the metadata graph. But 
> there are other ways to describe it using a URI.
>
> But it's not something to worry about for your list, since it's 
> trivial to support it with the system I've been advocating. It just 
> requires a convention to assign a URI to the file.
You got me! You are right, in the context of Metadata may exist Metadata 
about Metadata. I have to be more careful in the future, how I describe 
circumstances.
What I actually meant was that it only make sense to use the upcoming 
Metadata mechanism - whatever it will be - solely on the Content but not 
on Styles nor recursively on the Metadata.
>
>>     ◦     Styles are the formatting of Content and therefore have a 
>> relation in ODF to Content. Metadata as well will have a relation in 
>> ODF to Content. But only as Metadata and Styles often correlate, does 
>> not mean they have a direct semantic relation. They only have an 
>> indirect relation over the same content they are referring to.
>>  (There might be the idea of adding an attribute to a style to 
>> express it's consistent relation to a certain Metadata, but this 
>> won't need the same generic mechanism to relate multiple Metadata to 
>> content and is therefore neglected. Moreover it is redundant 
>> information).
>>     ◦     Metadata about Metadata might be a written as a new 
>> instance of Metadata and can therefore be neglected as well.
>
> "Neglected" in the sense that it doesn't matter to this particular 
> focus (associating document content with metadata).
Yep.
>
>>     4     A "Document Object" is either the Document itself or needs 
>> to have 'user data' as descendent (XPath axes)
>> User data can be any text or binary data the user added to the 
>> document aside of Styles and MetaData.
>>  Any comments, before I try my XSLT skills on this?
>
> The only thing I would say is be careful about thinking too much about 
> xpath for this.
XPath are very useful to specify an algorithm to find the "Document 
Objects" to be implemented in XSLT. Why do you think it is dangerous?
>
> For now let's just worry about what document objects we need to 
> support making metadata statements about.

No need to worry. ;-)
Furthermore, I have an addendum to the subset of XML elements:
We may divide the subset into two subclasses of XML elements. One 
directly containing user data (binary & text data), the other the 
ancestors of them.

More comments on this from you or one of the others?
Any comment - even declining - is most welcome!!

Best regards,
Svante