On the usage of xml:id and about (was: Re: [office-metadata] Contentduplication and ODF related RDF vocabulary)

[Elias Torres <eliast@us.ibm.com>]

Since there's a lot of confusion on xml:id and about, I'll try to explain
more what they are and how they are used in RDF/RDFa.

"Bruce D'Arcus" <bdarcus@gmail.com> wrote on 12/18/2006 08:13:27 PM:

[snip]
>
> But it would be a way to associate content with the metadata and
> address your expressed interest in using the same URI on multiple
> nodes, while having a single metadata description in the package.

You would use xml:id just to establish the URI of something represented by
that page but it doesn't mean its content represents it or the content is
attached to that resource in any way. For example:

Inside some index.html page:

<a name="elias"></a>

What does that mean?

Does the empty content represent me? No.

However, I can always in an external file have a triple:

<index.html#elias> foaf:homePage <http://torrez.us> .

Now for a more complex example:

Inside cameras.html

<ul>
  <li id="nikon_d200" about="#nikon_d200">
    <span property="dc:title" type="xsd:string">
      Nikon D200
    </span>
    purchased on
    <span property="dc:date" type="xsd:date">
      2004-06-01
    </span>
  </li>

  <li id="canon_sd550" about="#canon_sd550">
    <span property="dc:title" type="xsd:string">
      Canon Powershot SD550
    </span>
    purchased on
    <span property="dc:date" type="xsd:date">
      2005-08-01
    </span>
  </li>
</ul>

id is equivalent to our use of xml:id in the example above. However, if you
notice id by itself doesn't really mean much. It's just a placeholder for a
resource/URI. Because without it, there would not be a subject/resource
<http://example.net/index.html#nikon_d200>. However, the metadata about
that resource is expressed with properties inside the span and the about
attribute on each li element. In other words, it's not clear to me that
attaching an xml:id means anything except for establishing the location of
some resource. Any way we want to get at the content of the xml:id would
have to defined by us in our specification. RDF doesn't do that for us.
RDFa does do that for HTML through the use of about, property, etc.

Also notice that the purpose of usind xml:id is to say more things about
that subject on the same or other pages.

<ul>
  <li about="23456"> <img src="/user/markb/photo/23456.jpg" />
    ...
    using the <a rel="shutr:takenWith"
         href="cameras.html#nikon_d200">Nikon D200</a>,
    ...
  </li>
...
</ul>

Now we can say that pictures 23456 was taken with camera
<cameras.html#nikon_200>. Notice, how we are not interested in the "blob"
of HTML below it by itself.

>
> But to go back to the above, what are you trying to model here? What
> are the triples?
>
> In my example, if we assume this in the package:
>
> <rdf:Description rdf:about="http://ex.net/foo";>
>    <rdf:value>foo</rdf:value>
> </rdf:Description>
>
> ... then by way of the about attribute, we can say of that span:
>
>    <http://ex.net/foo> rdf:value "foo" .
>
> A quite simple statement, but nevertheless.

We cannot say that really (I think). meta:about is to establish a subject
anywhere in the tree, by no means it's attached to the current element. Let
me exemplify.

A page located on http://www.ibm.com/

<div about="http://torrez.us/who#elias";>

  ... a whole bunch of html here

  <div property="worksAt">
      IBM
  </div>

  ... a whole bunch of html here

  <div property="hasKids">
      yes
  </div>

</div>

Would you then say that http://torrez.us/who#elias rdf:value "... a whole
bunch of html here <div ... </div> ...." ?

>
> Bruce
>
>