Public identifiers and topic maps

W. Eliot Kimber eliot at
Mon Sep 28 19:19:46 BST 1998

At 05:15 PM 9/30/98 -0500, Steven R. Newcomb wrote:
>[Eliot Kimber:]
>> I think there's two different things being talked about here:
>[ ... and lots of other good stuff with which I agree.]
>But, Eliot, your note does not address the problem we're trying to 
>solve here.

In trying to respond in depth to Steve's note, I realized what the
fundamental problem is. Steve is talking about topics as though the topics
were the things (i.e., the topic "Lake Geneva" *is* the actual lake).  But
topics are not the things, they are descriptions of and opinions about
things.  That's why I say that a topic is a document.

We can prove that there exists in the alps of Europe a body of water that
has some measurable position on the globe.  That is a fact.

As soon as we start saying things like "this body of water is a lake" or
"this lake is called 'Lake Geneva'" we have asserted opinions about this
pool of water.  The opinion is not the thing.  The opinion points to the
thing.  Topics are just formalized forms of these types of statements.
They are abstract ideas that have to be documented if communication about
them over any useful time scale is to occur.

Why do we know that "Lake Geneva" is called "Lake Geneva"? Because someone
somewhere wrote it down the assertion: this body of water to which I refer
is called, by this group of people, "Lake Geneva".  They created the topic
"Lake Geneva" by writing down the assertion that the body of water is
called "Lake Geneva" by at least one person.  The topic is the idea that
this body of water is called Lake Geneva, not the body of water. The
document that says this is one member resource of the topic.

The names of topics are not and cannot be distinguishing, in the general
case, such that you can tell from two topic names whether or not the topics
are the same or different. I can create a topic called "Lake Geneva" by
which I mean all lakes called "Geneva" anywhere in the world. The only way
you can distinguish my topic from Steve's topic is to find all the members
of each topic and compare them for identity.  By same token, I can create a
topic with the name "That Lake in Switzerland" that is identical to Steve's
topic named "Lake Geneva" (identical because it includes the same member

It has always been and will always be the case that if two names are the
same (within the same name space, of course) then they must refer to the
same resource. But the converse can never be proved: if two names are
different, there's no guarantee, in the general case, that they don't refer
to the same thing.

You could define a name space in which you impose the rule that every
resource shall have exactly one name, but then you have the problem of
defining identity of resources.  For things like printed books or human
beings, it's relatively easy because they have lots of inherently
distinguishing properties, like author, title, publisher's ISBN number,
fingerprints, unique location in space and time, etc., that make it easier
to inspect names to see if they might actually refer to the same thing.
But topics, being more abstract (they're just ideas and opinions with no
well-defined physical or electronic representation), don't have inherent
distinguishing properties, so you can't use them when constructing names.
It would be up to some topic cataloging service to determine when two names
really referred to the same topic and disallow the cataloging of the second
name.  But this is a function of the catalogers, not the naming mechanism.


<Address HyTime=bibloc>
W. Eliot Kimber, Senior Consulting SGML Engineer
ISOGEN International Corp.
2200 N. Lamar St., Suite 230, Dallas, TX 75202.  214.953.0004

xml-dev: A list for W3C XML Developers. To post, mailto:xml-dev at
Archived as:
To (un)subscribe, mailto:majordomo at the following message;
(un)subscribe xml-dev
To subscribe to the digests, mailto:majordomo at the following message;
subscribe xml-dev-digest
List coordinator, Henry Rzepa (mailto:rzepa at

More information about the Xml-dev mailing list