RE: URI for language identifiers

> -----Original Message-----
> From: ext Peter F. Patel-Schneider 
> [mailto:pfps@research.bell-labs.com]
> Sent: 03 April, 2003 15:11
> To: Stickler Patrick (NMP/Tampere)
> Cc: www-rdf-interest@w3.org
> Subject: Re: URI for language identifiers
> 
> 
> 
> [I'm trying to get this thread back on a single track, so I'm 
> responding to
> this particular message, as much of the previous messages 
> (mine included)
> are getting off on tangents.]
> 
> From: <Patrick.Stickler@nokia.com>
> Subject: RE: URI for language identifiers
> Date: Thu, 3 Apr 2003 10:27:37 +0300
> 
> > > Oh agreed, the US has a very expansive notion of property.  
> > > However, even
> > > in this very expansive notion of property there are (still) 
> > > considerable
> > > ``fair use'' provisions.  Hopefully these provisions will 
> not be so
> > > weakened that I will be prohibited from making the claim that the
> > > denotation of http://www.whitehouse.gov/#43 is Tipper 
> Gore's husband.
> > 
> > I would hope so.
> > 
> > Patrick
> 
> But it appears to me that this is precisely what you are 
> trying to prohibit
> through specifications instead of legal means.

Again, as I've tried to clarify several times, I see two distinct
issues here:

1. The determination of ownership of a URI
2. The rights of an owner of a URI to specify denotation.

The first issue can involve legal machinery, the latter need not.

One can view this first issue as a determination of who has
the right to create a particular URI, insofar as that URI intersects with
matters of copyright, property rights, trademark law, domain ownership, etc.

A reasonable basis for URI ownership is that the creator of a URI is the owner, 
insofar as they have the legal right to mint that particular URI; or if the
creator acts as an agent of some other entity, such as an employee of a company 
where that the rights to mint a given URI are bound to their role as an employee, 
the owner may (though not necessarily) be the company.

But once the first issue is resolved, the second should be quite clear
and easily specified in technical standards.

I.e. insofar as the creator has the right to mint a given URI, they have the
right to specify its denotation.

Only the determination of the right to mint a given URI, governed by various
legal and social constraints, represents a potential challenge.

Presuming ownership is uncontested (which I think for most http: and other URIs
is the case, based on existing technical and social conventions based on domain
name and web server organization) the rights of the owner to specify denotation
should be crystal clear.

> I do not see any normative aspect of the World Wide Web or 
> the Semantic Web
> that provides any indication that there is a denotation 
> authority for URI
> references.  

Explicitly, no. And I hope this is addressed as soon as possible.

However, I believe that the concept of a URI owner and right of
a URI owner to specify the denotation of a URI is implicit in 
existing normative specifications as well as reflected by a common
view of users and architects, so the absence of an explicit
definition does not mean it does not exist and can thus be 
completely dismissed.

> I do not see any normative aspect of the World 
> Wide Web or the
> Semantic Web that mandates a single denotation for every URI 
> reference.

If you read the RDF specs in total, you will find that there is
a clear concept of unique, consistent denotation of URIs. Even though
the MT does not spell this out explicitly (which I personally consider
to be a flaw and shortcoming in the MT) the specification for the merging
of RDF graphs clearly reflects the premise that a particular URI has
globally unique and consistent denotation and that whenever a given URI is 
encountered in any arbitrary RDF graph, it always denotes the same
resource -- since any two nodes in any two RDF graphs having the same
URI are always merged to the same node when the graphs are merged. If 
that doesn't reflect a strong view of unambiguous denotation, I don't know
what would.

Furthermore, while the MT does in fact allow for different interpretations
(MTs seem to be slippery animals) there is very strong concensus that
non-monotonicity is a very, very bad thing and should be guarded against,
and that any interpretation of a given URI at a lower level of the SW
architecture should be maintained consistently at all higher levels of
the SW architecture. I.e., the denotation of a given URI should be 
globally unambiguous and consistent througout all interpretations at all
levels of the SW architecture. 

The RDF specs do not say this, because this is outside the scope of RDF.
The RDF specs cannot say how higher layers of the architecture must behave.
This is an issue to be specified for the whole of the SW architecture.
RDF only defines one level/layer of the architecture. Hopefully there will
be some expression of this in a forthcoming normative specification or
note from the SW Coordination group or the TAG.

But the RDF specs clearly reflect this position. To argue that they do not
is both incorrect and misleading.
 
> I do see many indications of how one might build such a denotation
> authority, if this was thought to be a good idea.  I also see many
> indications of how one might build a theory of denotation for 
> certain kinds
> of URI references (notably excluding http URI references with fragment
> identifiers as they are used in RDF).
>
> However, I think that it is a bad idea to base the entirety 
> of denotation
> in the Semantic Web on such principles.  I strongly believe 
> that this would
> serve as a bar to using the Semantic Web for many purposes.

Can you provide specific examples which clearly demonstrate how ambiguity
of denotation is a good thing, that it is required for some clearly useful
functionality? I've yet to see a single motivating example of how
ambiguity or non-monotonicity is beneficial to a formal system aimed
at the precise expression of knowlege. It would be quite something to
see one.

Patrick

Received on Friday, 4 April 2003 01:19:36 UTC