On Feb 5, 2007, at 4:00 PM, Xiaoshu Wang wrote:
The lie would be if you did a geturl on http://purl.example.com/
#aColor and retrieved the bytes "010203" and you did a get on
http://foo.com/#aColor and retrieved the bytes "030201". Remember,
you said they were sameAs, and same information resources have the
same bytes. That was the point. Since this is the case, it doesn't
matter if you retrieve one or the other of them, which was
justification for saying that you may try any getMethod and stop
after the first one that retrieves something.
I'll point out once again the confusion between
NotAnInformationResource and InformationResource. Colors are not
information resources.
The owl:sameAs is explained by the semantics of RDF model. Not byte-
by-byte. The same RDF model can be coded in a great many ways in
RDF/XML as well as with various flavors of n-3. On the other hand,
the the same-byte stream given to you can mean different thing as
well. For instance, if the same byte of an RDF document passed to
you with a MIME type of image/jpeg, you can certainly viewed it as
an image. I think you have probably messed up here.
Define identity however you want. I did and documented it. This
matter will not be settled with reference to the semantics of the RDF
model. The RDF model does the best thing it can possibly do based on
what it has access to: "... making an assertion amounts to claiming
that the world is an interpretation which assigns the value true to
the assertion".[1] Defining identity of resources is a necessary
step if you are to make this claim.
However you define identity, the point remains the same. If you say
that two URIs denote the same thing, then retrieve them and find that
your chosen definition of identity doesn't say they are the same,
then you have "lied" in making the sameAs statement. Failing that you
may use whatever the first successful getMethod retrieves.
To make what you have proposed work, you must cleanly separate the
resources that you are discussing. That means, you need another
another URI (let's dub it URIs) that points to the "URI" (let's dub
it URIr) pointing to the resource of your interest. Then, you can
say that the all URIs is a string. Then, perhaps, you can describe
them more meaningfully. But again, you need to define URIss, and
URIsss, etc...It reminds me of what Godel's incomplete theorem has
told us - Dont't waste our time on designing a system that can do
everything because there isn't any! You will always end up with
more, and often harder, questions than what you have before. URI
is the foundation of RDF because each node and edges are defined by
a URI. And now you are trying to define node of node, nodes of
arc, arc of arc, arc of nodes. And you won't have a complete
solution, period.
You are going to have to be more careful than this if you are going
to convince me of anything. Carefully set up a problem then show how
our proposal leads to consequences that are wrong in some way. Godel
set up his problem statement and proof very carefully - perhaps thats
where the lesson should be.
I remember we discussed this problem before. And my point was that
the RDF world is an open world. Therefore, given one URI, you need
to have a solution that can promise the knowledge of all possible
URI's resolution.
The "therefore" doesn't follow from the antecedent in any way that I
can discern.
If you cann't, then you will always end up with some 404, then you
are back to the problem. If you can, tell me this chunk knowledge
won't be huge? And how you are going to syn the knowledge on all
machines? (It doesn't matter if you call it DNS or not, it would be
something similar). You honestly think this can work? I really don't.
OK. Perhaps we should leave it at that.
Best,
Alan
[1] http://www.w3.org/TR/rdf-mt/#entail paragraph 2