[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: The Map/Territory Conundrum in Topic Maps vs. RDF
- From: "W. E. Perry" <firstname.lastname@example.org>
- To: XML DEV <email@example.com>
- Date: Sun, 19 Aug 2001 23:19:29 -0400
Jonathan Robie wrote:
> At 12:09 PM 8/19/2001 -0400, W. E. Perry wrote:
> >May I respectfully suggest that in the inherent nature of markup the
> >map is in fact--in a real and inalienable sense--the territory, and vice
> >versa, of course.
> I don't think this is true.
> In my talk, I showed an example with Rembrandt's "Portrait of the artist at
> his easel". The RDF assertions I used showed that the resource identified
> by http://www.artchive.com/rembrandt/artist_at_his_easel.jpg is has the
> mime type "image/jpeg", and it hangs in the Louvre.
> I rather suspect that the "Portrait of the artist at his easel" that hangs
> in the Louvre does not have a mime type, but the "Portrait of the artist at
> his easel" found on http://www.artchive.com/ does. These are two quite
> different things. When we mix up the map and the territory, we wind up
> creating silly worlds in which an oil painting has a mime type.
Between this last paragraph and the previous one, you seem to me to have
answered the question. As you point out in the latter paragraph, what hangs in
the Louvre does not have a mime type. Therefore markup which provides some
gloss on content characterized as of type "image/jpeg" is a 'map' to that .jpg
reproduction and not to the painting itself. As you know (as well as only an
expert can) the question of how to make assertions in markup about the physical
painting, rather than about some textual representation of or commentary upon
it, is the most contentious single question in ontological mapping.
The crux of the problem is that markup and the content upon which it comments
are, in the first instance, text and are processed first with lexically
focussed tools. If particular content encodes an image, markup specific to that
content is, at best, applicable to the image, but not to any artifact which,
though pictured in that image, has or had an existence independent of that
image. This is inherent in the nature of the medium and is surmountable only
through indirection and tacit agreement on the intent of that indirection. The
stuff of markup and of the content of marked up instances is text. If the name
of the painting--"Portrait of the artist at his easel"--appears as content and
is the subject of markup, that markup is literally specific to that title and
not to the painting thus identified. The audience for such markup is generally
expected to be cognizant of the indirection through which assertions literally
about the title of the painting are for the most part understood to be
assertions about the painting itself. But strictly speaking, because the marked
up content is itself text, it is that text which the markup maps or otherwise
comments upon. Ontologically this is precisely equivalent to saying that the
painting whose subject is Rembrandt is not itself Rembrandt, because of the
interposition of the medium--in the case of the painting, oil and canvas--which
is its physical content in precisely the same way that the content of the
marked up instance is text.
> This is an important issue for distributed authoring of RDF: how do people
> know whether they are making assertions about the same thing?
> I wish I hadn't been in such a hurry on Friday, Walter. We could have had a
> long and pleasant breakfast discussing these things.
I agree. Merely deferred, I hope.