Re: Documents, Cars, Hills, and Valleys

[GOWADIA, VAIBHAV ]
>
> This is my first posting in this group, so let me introduce myself.

Good, an injection of fresh blood!

> > From: Thomas B. Passin
> > In this particular example, it's easy to assume that the subject of
> > http://www.microsoft.com is in fact Microsoft the company
> > (but how would a
> > computer know this if a person had not already encoded it?).
> > Once we have
> > the subject, we do not have to refer to it as
> > http://www.microsoft.com. We
> > can remove the ambiguity for users of our triples about it.
> > For example,
> > we can create an anonymous node and say that its type is
> > "company" and that
> > the uri representing this resource is (for example)
> > "urx:companies:microsoft", then use that uri.
>
> There can be more than one ways to identify a particular resource.
> Thus if we use URI's of type 'urx:companies:microsoft' then how do we
> distinguish between following resources:
> urx:people:rdf-interest:Vaibhav
> urx:people:java-interest:Vaibhav
> In such a case how can we verify whether these two persons are same
> or different ?
>

It may not be possible to know if we do not have more information.  After
all, if you refer to "T. B. Passin" and someone else refers to "Tom P", you
may not know that they are talking about the same person (and perhaps they
are not).  But at some point, you may discover some statement that
definitively shows that they are the same person (perhaps they both turn out
to have the same Social Security number, for Americans).  At that time, you
can add some kind of equivalence statement.  In Topic maps, you would merge
the topics into one.

> We should not try to force a unique URI for all things that would be
> described on the web. Otherwise, if one wanted to describe
> http://www.cse.sc.edu/ (department, not website), the person may
> first have to find a URI for the resource using a
> "Resource-URI-Directory-Service".
>

I don't think the issue, for Semantic Web-like consideration, that one thing
should have one unique URI, convenient though that might seem to be.  I
think we want to be sure that one URI does not indicate more than one thing.
If we can get it to be unique, fine, so much the better.

> One more point that needs to be considered in designing a URI syntax
> is: How to write a URI for an imaginary resource ?
>

urixxx:tbp:another-figment-of-my-imagination

There, I just wrote one.  Can I convey its meaning, though? Harder.  Can I
convey its meaning without human-to-human explanations in natural language?
Not only is that probably harder, but it gets into the question of its
"meaning".  That hasn't been settled in over 2000 years, not really.
Probably the best we can hope for is some widely but loosely understood
operational definition.  And who is to say that that's not how people come
to understand a lot of things, anyway?

Or maybe that's not what we need from RDF or the Semantic Web at all.  Just
like a lot of theorems can be proved withoug any understanding of what they
"mean", and a lot of information retrieved without knowing its "meaning",
maybe we can do most of what we want without ever knowing the exact
"meaning" of urixxx:another-figment-of-my-imagination.

>...
> To summarize:
> 1.A URI may not be unique text string but all URI's about the same
> resource should map to same resource.

Nice, but if we need a central repository to accomplish the mapping it's not
going to work out for the Web.  Some other way would be needed.  Also, the
prescription is bound to be violated in practice so we still have to handle
the case of bad or missing mapping, just like we do now.

Cheers,

Tom P

Received on Thursday, 25 April 2002 00:02:29 UTC