RE: Xinclude word-smithing

Hi Ian,

> From: Ian Davis [mailto:Ian.Davis@talis.com] 
>
> > Point of clarification: it is not just a matter of producing 
> > fewer triples, it is also a matter of producing *different* 
> > triples -- different assertions.
> > 
> Can you give a concrete example of this? Bear in mind that:
> 
> 1) The author of the document has free choice of the GRDDL
> transformation to apply
> 
> 2) The author of the document decides whether to use xInclude 
> markup or not

Yes, but this choice may already have been made -- years ago -- or it
may be otherwise outside the control of the GRDDL transformation author.
Other software may dictate the format, for example.  We can assume that
the GRDDL transformation author *understands* the XML document format,
but we cannot assume that he/she has any control over it.

A key application of GRDDL that I am concerned about is where GRDDL is
used to treat arbitrary XML documents as custom serializations of RDF.
If by separate policy the resulting RDF is treated as authoratiative
regarding the meaning of the XML document, then:

 - The same XML document can be used by both XML and RDF applications
and have the same meaning.

 - The syntax of the XML document can vary (across versions, for
example), while consistent meaning is retained.

Here is an example, based on the one given in my initial issue-dbooth-3
comment.  You can imagine that this example is a distilled version of a
situation where XML documents contain chunks of other XML documents.  

<myns:myDoc . . . >
   <myns:quote>
      <otherNs:whatever>
         <xi:include href="http://example.org/do-not-expand" />
      </otherNx:whatever>
   <myns:quote>
</myns:myDoc>

In this example, everything inside the <myns:quote> element is intended
to be effectively quoted, i.e., the <myns:quote> element acts as an
envelope for what is inside.  Given this XML document, I wish to write a
GRDDL transformation that produces exactly the following triple:

	myns:myDoc   myns:quote  '<otherNs:whatever>
         <xi:include href="http://example.org/do-not-expand" />
      </otherNs:whatever>'.

I.e., the value of the myns:quote property is supposed to be the
syntactic XML chunk that was inside the <myns:quote> element.  If
XInclude expansion is not done, then I can get the right answer.  But if
XInclude exapansion *is* done, then assuming that deferencing the URI
http://example.org/do-not-expand yields the word 'apple', I instead get
a triple like:

	myns:myDoc   myns:quote  'apple'.

Thus, as GRDDL transformation author, if I cannot control whether the
XInclude is processed, then I cannot be unambiguous in denoting the
intended RDF results.


David Booth, Ph.D.
HP Software
+1 617 629 8881 office  |  dbooth@hp.com
http://www.hp.com/go/software

Opinions expressed herein are those of the author and do not represent
the official views of HP unless explicitly stated otherwise.

Received on Sunday, 17 June 2007 16:36:22 UTC