Re: [office-metadata] Our discussion on the Wiki example

[Bernd Schuster <schuster@mskgmbh.com>]

Hi
first of all, I  wish all of you a happy new year !

Bruce, Michael,
for more clarity from my side a few comments about this thread:

Bruce D'Arcus schrieb:
>
> On Dec 28, 2006, at 12:28 PM, Michael Brauer wrote:
>
>> Hi,
>>
>> Bruce D'Arcus wrote:
>>> Hi Bernd,
>>> On Dec 28, 2006, at 7:02 AM, Bernd Schuster wrote:
>>>> From my point of you, there are no objections to write:
>>>>
>>>> <text:p xml:id="myBook">
>>>> My favorite books is from Tolkien! It's ISBM is <text:span 
>>>> xml:id="isbn">8th</text:span> it has <text:span 
>>>> xml:id="pages">1154</text:span> pages
>>>> </text:p>
>>>>
>>>> ... and in the RDF/XML:
>>>>
>>>> <rdf:Description rdf:about="content.xml#myBook">
>>>>  <ex:author rdf:resource="http://ex.net/people/Tolkien"/>
>>>>  <ex:isbn rdf:resource="content.xml#isbn"/>
>>>>  <ex:pages rdf:resource="content.xml#pages"/>
>>>> </rdf:Description>
>>>>
>>>> where the "words" '8th' and '1154' are  tagged  with an individuell 
>>>> xml:id  and being referenced  as objects inside the above rdf 
>>>> statements.
>>> Yes, the objections are:
>>> 1) it is not a standard way to model RDF, and would in fact violate 
>>> many ontologies. In short, we would force every literal to be a 
>>> resource.
>>> 2) it forces additional processing, and the metadata is now 
>>> dependent on the (ODF) content
>>
>> I'm wondering whether the "1154" that appears in the content.xml 
>> actually is an RDF object, or only the display string of an object 
>> that is stored in a RDF file external to the content.xml.
>
> I was assuming that string is the literal object (if it wasn't, then 
> you'd need to duplciate the literal in the RDF/XML), but I guess you'd 
> need to be explicit about defining the processing expectations.
In fact, my idea was, that both subjects and objects might be stored 
inside the content.xml and being referenced out of  metadata files (as 
shown above). But this is only a conception. Why not storing rdf 
subjects and/or objects inside meta data files while referencing them 
from the content.xml, if this is acceptable from the engineering side ? 
Maybe this is a solution to avoid duplication of data.
>
>> In this example, I would assume that all bibliographic data is stored 
>> externally, and that the content.xml only displays some of the RDF 
>> objects. What we would need in this case is not a reference from the 
>> RDF data to the content.xml, but a reference from the content.xml 
>> back to the bibliographic meta data, so that the display string can 
>> be updated if the bibliographic meta data changes.
>
> In general for this use case, yes. But Berndt (as I say below) might 
> have something else in mind.
>
>> That's actually how the meta data fields we have already in the ODF 
>> spec do work. They do not define any meta data, but display it only.
>
> I think this is a useful distinction you make, and one we will need to 
> clarify: content *as* metadata, vs. fields that use metadata for 
> display (or other). Elias spreadsheet example is the first, and my 
> citations are the second.
>
> Berndt's here is a little unclear actually; am not sure what he intends.
Fields, that show meta data were not explicitly in my mind, while 
writing down the above example. My focus was on subjects and objects 
which are part of the human readable content of the document, such like 
words, phrases, table fields or paragraphs.

>
>> If my assumption is not correct and the "1154" in the content.xml is 
>> an RDF object, wouldn't this mean that the content.xml defines 
>> bibliographic data itself? How would this data be kept up to date?
>>
>> Are my assumptions correct, or do I miss something?
>
> I think you ask the right questions. If we he wants it to be live 
> metadata, then its right there, in the content.
Both scenarios are possible:
1) content is tagged with meta data and changes during several editing 
sessions before it is "finalized" (and therefore must several times be 
updated whereever it is stored (content side or meta data side). I hope 
this is what you call "live metadata".

2) content is frozen and tagged afterwards with meta data for example 
during archiving.
> But I could also imagine him using a field to display the data pulled 
> in from the package instead.
That's right. From my point of view this is rather an engineering 
question than a conceptual one.

Bernd
>
> Bruce


-- 
mit freundlichem Gruß

Bernd Schuster

MSK Gesellschaft für Automatisierung mbH
Friedrich-Ebert-Allee 33
D-22869 Schenefeld

Tel.: +49 40 839286-10
Fax : +49 40 839286-29

eMail: schuster@mskgmbh.com