RE: designating datatypes

 

-----Original Message-----
From: ext pat hayes [mailto:phayes@ai.uwf.edu]
Sent: 04 March, 2003 12:00
To: Stickler Patrick (NMP/Tampere)
Cc: w3c-rdfcore-wg@w3.org
Subject: RE: designating datatypes



Pat,

Person names are not globally unique. That's why
we have identifiers such as social security numbers,so that we have globally and temporally unique names.

URIs are intended to be globally and temporally unique
names.


The names are unique in the sense that they are proof against ambiguity about which name is which (unlike personal names, eg there are hundreds of people called "Pat Hayes").  They are not unique in the sense that they have a unique referent.


 
 
If you mean by 'unique referent' that they only denote one thing, then I strongly
assert that they do (even if the specs or MT don't say it clearly enough).
 


SS numbers (barring occasional bugs in the system) do have a unique referent, but the machinery which provides that uniqueness is elaborate and socially embedded,  requires a large degree of cooperation between many people over a long time period, and is influenced by national laws, for example.  If this social superstructure were not in place or were not operational, then SS numbers would not have unique referents. There are NO such mechanisms for URIrefs; the issue isn't even discussed in the relevant specs.

The closest that RFC 2396 comes is this:

"Resource
A resource can be anything that has identity.  ... Not all resources are network "retrievable"; e.g., human beings, corporations, and bound books in a library can also be considered resources.


The resource is the conceptual mapping to an entity or set of entities, not necessarily the entity which corresponds to that mapping at any particular instance in time.

Identifier
An identifier is an object that can act as a reference to something that has identity.  ...


Having identified a resource, a system may perform a variety of operations on the resource .... "

Notice the non-sequiteur at the end of this, where we have magically moved from 'can act as a reference' to being a means by which one can 'identify' the referent, and then even more mysteriously, having identified it, one can perform operations on it. To see what is wrong with this, consider the fact that this sentence refers to Alpha Centauri, and I think I can say without risk of being thought overconfident, actually manages to refer to said astronomical body; but I wouldn't expect the ability to refer to a star would give me the ability to perform any operations on it. If I were to say that Julius Caesar probably saw Alpha Centauri, then Ive referred to something that I certainly cannot perform any operations on, since his atoms have long since been spread thinly over the entire planet.

 
Well, 2396 says
 
[
   Having identified a resource, a system may perform a variety of
   operations on the resource, as might be characterized by such words
   as `access', `update', `replace', or `find attributes'.
]
 
I would expect one could always 'find attributes' about a resource, and 2396 doesn't
say that the same 'operations' must be applicable to all resources.

 
If the RDF MT does not reflect that, then it is IMO
fatally flawed.


Its not a flaw to fail to do the impossible. 

We seem to just be disconnecting here. If it is impossible to presume that a given
URI always denotes the same thing wherever it is encounted in RDF-space, then
what use is RDF for reasoning...? 


Let's discuss this in Boston... (hopefully you're
attending the plenary)


I'm here right now :-)

Pat

 
I just got in, and am jet lagged, so I guess we'll meet up Wednesday.
 
Cheers,
 
Patrick