Re: [office-metadata] Finding a common proposal..

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

On Dec 5, 2006, at 7:40 PM, Svante Schubert wrote:

> Of course you can say that this area is a semantic black box handled 
> by the implementation of the plug-in.

Yes, that's what I would say.

> But as there might be many plug-ins and not always available those 
> areas become a none-edit field.

Yes, though I suppose as in Word with the option with flattening them 
(converting to editable text).

This isn't a new problem that we're adding, BTW. It's a general issue 
with custom fields. I really don't think it's our problem, but the job 
of GUI designers and such to manage.

> Editing these fields bring you into danger to become inconsistent.

OK.

> On the other hand, if you define "Doe" and "1999" as further areas of 
> meta data, it would enable you to keep them consistent with any 
> reference you have.
> Therefore consistency is the requirement you asked for.

Remember, the use case is pretty specific that a citation's 
presentation can radically change, including not only the specific 
properties included, but the order of the resources, whether they are 
presented in-text or in footnotes, etc., etc.

E.g. the same citation can be presented like:

	(Doe, 1999; Smith, 2000a:23)
	(Smith, 2000a, p23; Doe, 1999)
	[1]
	(1) footnote mark, with full references in the footnote text	
.... and any number of variants. User needs to be able to change 
between them in real-time, without modifying the source.

Your proposal doesn't solve that problem I'm afraid. Those metadata 
properties (in your example the authors' family names, and the year of 
publication, plus a year disambiguation suffix that can only be known 
by processing the entire in-document list of citation references; it's 
not intrinsic to the bibliographic metadata) are only really relevant 
to one kind of presentation style. They're pretty much irrelevant to 
last two, and in the footnote example, you need much more metadata to 
keep things -- as you say -- consistent.

Nevermind RDF and such; just go back to BibTeX. This proposal just uses 
the same approach, but updated for the 21st century. It's also the same 
basic approach MS is using in Word 2007/ OOXML (again, minus the RDF).

> BTW I found an argument for the design decision, that all viewable 
> content should remain in the content.xml. The reason: as multiple meta 
> data might like to refer to it, we would otherwise end up with 
> inconsistencies of data copies in various meta files as multiple 
> plug-in would handle the data independently.

In fact, one of the reasons I embarked on all this is because the 
existing ODF citation solution is fatally flawed in its design, 
precisely because metadata is added as attributes to each citation 
(text:bibliographic-mark). In a book where I might cite the same 
reference 50 times, that's a whole lot of redundant metadata.

Given all this, isn't it easier to just treat the presentation content 
as that black box?

Bruce