HTTP URIs need not be ambiguous, was Re: The case against URNs

Champion, Mike wrote:
>
> My assertion -- which I don't hold all that strongly, but would have to be
> persuaded to drop -- is that the "name" http://www.whitehouse.gov is
> fundamentally ambiguous -- it may be the site, or some abstraction -- and
> hence automated inferencing systems will always get confused easily when
> confronted with it.

 I am not sure what you mean by _fundamentally_ ambiguous. Suppose you were
to dereference the above URI: it _might_ say

"The website of the whitehouse"

or it might say

"A representation of the whitehouse:"

[picture of whitehouse goes here]

You might specifically ask for:

application/rdf+xml

and get back

<rdf:Description rdf:about="">
    <rdf:comment>The Whitehouse</rdf:comment>
    <rdf:type rdf:resource="...#GovernmentBuilding" />
    <rdf:type rdf:resource="...#House" />
    <ex:LivesIn rdf:resource="#President" />
    ...
</rdf:Description>

and then you'd have an idea that the URI references "the Whitehouse"

you might also get back some RDF which describes a directory of a website
and then you'd be able to draw different conclusions.

Where is the fundamental ambiguity? This really works, and we can use HTTP
URIs for practical things like Healthcare identifiers etc., and we can
develop software that implements OWL, and it works and draws helpful
conclusions about the data we enter into our little database.

Are you worried that it just won't work? RDF/OWL systems could care less
what the contents are of a string that forms the URI that they are using as
logical constants ... could care less.

Jonathan

Received on Sunday, 6 October 2002 20:56:25 UTC