RE: rdfms-resource-semantics

Patrick,

Without regard to the actual specifications for the various UR*s, are you
arguing that they *ought* to be used something like this:

--Pointers--

----Implicit pointers

(Implicit pointers can be de-referenced, and [in theory] the referent
retrieved.)

------Automated pointers
(The referent of an automated pointer can be retrieved by an automata.)

Examples of automated pointers.

char * p.   de-reference using *p

http://www.foo.bar/index.html.


By analogy a URL *should* indicate an automated pointer.  (With the proviso,
that like any pointer it might in fact be dangling or void when
de-referenced.)


------an-automated pointers

(A non-automated pointer is not guaranteed to be retrievable by an automata,
but by definition pointer _itself_ will--in principle--tell you how to
locate the referent.)

Examples:

You wisely held put derivative on the NASDAQ index in October 2000 and
decided to let it come due.  The entity referenced (a quantity of US e-$s)
appeared automatically in your brokerage account when the contract came due,
successfully hedging investments in technology stocks.

You held futures in barley due in January 2001.  Since you were going to
open a brewery you decided to take delivery on the contract.  Your contract
partner faxed your broker that your five metric tons of wheat were sitting
in the harbor at Rotterdam.  How you got it to New York was your @#$%
business.


In principle all URI's should be treated as an-automated implicit pointers.

----Logical pointers

(Logical pointers, unlike implicit pointers cannot be de-referenced without
explicit access to the translation table.)

Examples:

The literal "stupid"  you have to know that this must be de-referenced using

#! perl
my %stupid_hash{"stupid"};

Or

The RDF entry in a web page refers to the Library of Congress subject
category "G".
You have to have treat "G" as a logical reference and know that you will
look it up in the LC-Guide-To-WWW-Subject-Codes.  There you will see it
codes for "anthropology, general".


In principle all URN's should be treated as logical pointers.


---

RDF pointers are mostly names referring to internal categories; therefore
(in the main), RDF pointers should be URNs.

---------------

Is this a reasonable representation of your position?

Received on Saturday, 9 June 2001 21:58:15 UTC