W3C home > Mailing lists > Public > www-rdf-interest@w3.org > April 2003

Re: URI for language identifiers

From: Peter F. Patel-Schneider <pfps@research.bell-labs.com>
Date: Wed, 02 Apr 2003 08:42:33 -0500 (EST)
Message-Id: <20030402.084233.125099570.pfps@research.bell-labs.com>
To: Patrick.Stickler@nokia.com
Cc: miles@milessabin.com, www-rdf-interest@w3.org

From: <Patrick.Stickler@nokia.com>
Subject: RE: URI for language identifiers
Date: Wed, 2 Apr 2003 11:07:47 +0300

> 
> 
> > -----Original Message-----
> > From: ext Peter F. Patel-Schneider 
> > [mailto:pfps@research.bell-labs.com]
> > Sent: 01 April, 2003 17:20
> > To: Stickler Patrick (NMP/Tampere)
> > Cc: miles@milessabin.com; www-rdf-interest@w3.org
> > Subject: Re: URI for language identifiers
> > 
> > 
> > From: <Patrick.Stickler@nokia.com>
> > Subject: RE: URI for language identifiers
> > Date: Tue, 1 Apr 2003 16:32:09 +0300
> > 
> > > > -----Original Message-----
> > > > From: ext Miles Sabin [mailto:miles@milessabin.com]
> > > > Sent: 01 April, 2003 16:10
> > > > To: www-rdf-interest@w3.org
> > > > Subject: Re: URI for language identifiers
> > > > 
> > > > 
> > > > 
> > > > Patrick Stickler wrote,
> > > > > *How* or *where* a given URI is used does not affect its
> > > > > authoritative meaning. Usage can only reflect the presumptions
> > > > > of the user, but that does not usurp the authority fo the owner,
> > > > > and may very well result in disagreement or ambiguity.
> > > > 
> > > > I'm sorry, but this is just wrong.
> > > > 
> > > > Other than intrinsic network and DNS semantics, there 
> > isn't currently 
> > > > any notion of an authority for a URI which has semantic 
> > relevance to 
> > > > RDF.
> > > 
> > > Either you have not understood what I was saying, or you 
> > are simply wrong
> > > (or both ;-)
> > > 
> > > The denotation of a given URI is that which the owner of that URI
> > > specifies. Period.
> > 
> > Hmm.  Where does this come from?  Can you provide anything 
> > besides your own
> > gut feeling for this?
> 
> Can you provide anything to refute it other than your own gut
> feeling for this?

There are URI references for which there is no owner at all, such as
abcdefghiklmnopq:rstuvwxyz#a012345, so your claim breaks down for at least
these URI references. 

> However:
> 
> >From http://www.ietf.org/rfc/rfc2396.txt
> [
>    3.2. Authority Component
> 
>    Many URI schemes include a top hierarchical element for a naming
>    authority, such that the namespace defined by the remainder of the
>    URI is governed by that authority.  This authority component is
>    typically defined by an Internet-based server or a scheme-specific
>    registry of naming authorities.
> ]
> 
> where a "naming authority" is that agency which (either directly or 
> indirectly) mints a URI and defines its denotation -- i.e. which names 
> a resource by a URI.
>
> That authority either equates to, or may designate, the creator (owner) 
> of a given URI and it is the creator which specifies what the URI denotes.

Where in this is there anything that talks about the denotation of a
URI reference?  

This quote also mentions that governance for namespaces is only for ``many
URI schemes''.  What about the others?

> True, the actual agency is indirectly identified by the naming authority
> component of the URI scheme, being e.g. the owner of a given domain name,
> IP address, registry domain, etc. but the identity of the agency is
> clearly determinable.
> 
> And even if the determination of ownership for a given URI may take
> some investigation -- for any given user, it should be clear when
> they *don't* own a given URI, and thus should look to the owner for
> what it denotes. 

Again, what does naming governance have to do with denotation?  

> This issue has been explored at length by the RDF Core WG and (while
> not to be included in the final specs, but in a separate publication)
> the discussion regarding the meaning of URIs authored by the WG is
> illustrative of the shared view of many members of the SW community:
> 
> >From http://www.w3.org/TR/rdf-concepts/#section-Social:
> [
>    4.3 Authoritative Definition of Terms
> 
>    The social conventions surrounding use of RDF assume that any RDF URI 
>    reference gains its meaning from some defining individual, organization or 
>    context. This applies most notably to RDF predicate URI references. 
> 
>    These social conventions are rooted in the URI specification [URI] and 
>    registration procedures [URI-REG]. A URI scheme registration refers to a 
>    specification of the detailed syntax and interpretation for that scheme, 
>    from which the defining authority for a given URI may be deduced. In the 
>    case of http: URIs, the defining specification is the HTTP protocol 
>    specification [HTTP], which specifies how to use the HTTP protocol to obtain 
>    a resource representation from the host named in the URI; thus, the owner of 
>    the indicated DNS domain controls (observable aspects of) the URI's meaning. 
> 
>    Thus, the choice of terms used in published RDF is significant in determining 
>    its meaning, through reference to definitions asserted by the defining 
>    authorities for those terms. 
> ]
> 
> Thus, the views that I have been expressing here are both reflected in 
> the URI specs as well as reflect common opinion in the SW community.

A unanimous decision of the RDF Core WG
(http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2003Mar/0068.html)
struck the entire social meaning section from this document, so referring
to it doesn't carry much weight.  I haven't heard of any separate
publication from the RDF Core WG on this issue.  

> > > And any disagreement about the denotation of a given URI utimately
> > > must be resolved by the owner of that URI.
> > 
> > Hmm.  Where does the notion of the (unique) owner for (every) URI come
> > from?   For example, who is the owner of
> > 	http://www.bellllabs.com/user/pfps/#me
> 
> I couldn't say. Though to find out, I'd first contact the owner of 
> the domain www.belllabs.com, and I'd expect I'd likely be told that
> the sub-webspace http://www.belllabs.com/user/pfps/ was controlled
> by you, and that all URIs grounded in that sub-webspace are under
> your ownership. Therefore, you are probably the owner of the URI
> http://www.belllabs.com/user/pfps/#me and if I wanted an authoritative
> definition of what that URI denotes, I'd ask you specifically.
> 
> Is that really so difficult?

Yes, as www.bellllabs.com is not a registered domain.

> > Even if there is an owner of a URI, how does that owner 
> > determine what a
> > URI means?  
> 
> The issue is not *how* the denotation of a given URI is defined,
> but *who* has the right to define it.

Huh?  If there is no means for determining the denotation of a URI
reference, then I don't see how this scheme can possibly work.  Are you
proposing, for example, *the* denotation of a URI reference *could* be
determined by the thoughts of a dead individual?

> > Even if there is some platonic notion of the 
> > ultimate meaning
> > of a URI, how is that meaning conveyed to RDF (or DAML+OIL, or OWL, or
> > LBase, or any other formal system in the Web)?
> 
> You are now diverging to an entirely different issue.

I don't think so.  If there is no mechanism for conveying the denotation to
tools then what good is it?

> The three key points of this thread are:
> 
> 1. URIs should have unambiguous, consistent, global, and immutable denotation.
> 2. Each URI has a specific owner, who has the right to say what it denotes.
> 3. Statements made by owner of a resource can be considered to have special 
>    sigificance/authority over statements made by 3rd parties.
> 
> How the denotation of a given URI is defined, or how the semantics of the
> resource denoted are defined, are outside the scope of this particular
> discussion.

I disagree entirely.  If you can't turn your vision into something that
affects tools, then why should anyone care about it?

> They're important issues in their own right, of course. Feel free to
> start a separate thread to discuss them.

No thanks.  Theories of meaning in the Semantic Web should connect to the
formalisms of the semantic web.

> > > If I mint a URI and say it denotes the city of Paris, France. And
> > > you come along and use it to denote the city of Paris, Mississippi,
> > > then you are wrong and I am right, because it's *my* URI. 
> > 
> > How is it your URI?
> 
> Because I created it. Is that really so hard to understand?

Yes.  Even the notion of who creates a URI reference is extremely murky.
For example, who created the URI reference http://www.foo.ex/#a ?

> > > It's as
> > > simple as that. 
> > 
> > This is not simple, it is, instead, extraordinarily complex.
> 
> No. I think you simply want it to be complex. Or perhaps are
> not able to dissect a number of disjunct issues and address
> each individually. The issue of URI ownership and denotation
> is very simple.

So tell me who owns http://www.foo.ex/#a .
Tell me what http://www.foo.ex/#a denotes.

Tell me who owns http://www.dccomics.com/#clarkkent .
Tell me that http://www.dccomics.com/#clarkkent denotes.

> > For example, suppose DC Comics mints a URI reference, 
> > 	http://www.dccomics.com/#clarkkent
> > what exactly does this denote?  
> 
> Whatever DC Comics *says* it denotes. They created it,
> so they get to say what it means. Simple.

Not simple.  Suppose they don't say what it denotes.  Suppose I want to
point out problems with their use of the URI references.  Suppose their
entire world view doesn't make sense.

> If that denotation doesn't work for you, then don't use it.

Then how can I communicate with DC Comics, if only to discuss problems with
their view of the world.

> > Can I use this as a synonym for
> > 	http://www.dccomics.com/#superman
> 
> Probably not, since the names carry "perspective" in addition to
> denotation of some entity. But you digress.

Not at all.  You were proposing that there is some fixed
``unambiguous, consistent, global, and immutable denotation'' for a URI
reference.  If so, then synonym issues are yes/no issues.  Now you are
adding in some notion of ``perspective'' to fix problems with this rigid
designation view of the denotation of URI references.  I was simply
pointing out that denotation is a complex issue, and that simplistic
solutions quickly break down.

> Again, you are diverging into yet another issue that is completely
> disjunct from the issue being discussed.

I disagree.  You are proposing a solution for the denotation of URI
references.  I am pointing out problems with various aspects of this
solution.

> Whether or not two different names which denote the same entity can
> be used interchangably, as they may reflect distinct "perspectives" 
> of that entity (such as "Morning Star" and "Evening Star" both denoting
> the planet Venus) that has nothing whatsoever to do with whether a
> given URI has ownership and whether the owner has the ultimate right
> to say what it denotes.

Again I disagree.  

> Again, if you wish to explore other issues, feel free to start a
> new thread.

No thanks.

> > > The owner of a URI gets to say what it denotes and
> > > if you don't agree, then don't use that URI. Use of someone elses
> > > URI is an implicit agreement about what the owner says it denotes
> > > (leaving aside cases of misunderstanding, etc.)
> > 
> > So how then can I utilize URI references such as 
> > 	http://www.whitehouse.gov/#43
> > I may want to use this well-known URI as a common source of 
> > URI references
> > for the U. S. presidents, so that I can communicate with other people.
> > However, I, like Michael Moore, may disagree with the denotation that
> > www.whitehouse.gov gives to this particular URI reference.  
> 
> Tough. Too bad. If you don't agree with, or can't use the authoritative
> denotation, then you'll have to find some other names to use.
> 
> Actually, I don't see how you can "disagree" with what the creator
> of a URI says it denotes, since you have no authority to say otherwise.

In your view of denotation, I can't.  However, I don't agree with your view.

> You may disagree with some statement made about the *resource* denoted
> by a give URI, but you don't have any basis for disagreeing with 
> the denotation specified by the creator of the URI, as the creator of
> a URI has the sole right to say what it denotes.

Again, only in your view of the denotation of URI references.

> > If I have to
> > use the denotation given by www.whitehouse.gov, then I have 
> > no way of even
> > expressing disagreement with that denotation.
> 
> Correct. If you use that URI, you implicitly agree with its specified
> denotation. If you don't want to talk about what that URI is specified
> to denote, then don't use that URI. Pretty simple really.

Well, sure, this is a consequence of your view of the denotation of URI
references.  I view this as a severe problem with your view, and believe
that a different view of the denotation of URI references is needed for the
Semantic Web.

> > > Likewise, if I have a URI that denotes a document that I wrote and
> > > of which I am the owner, and I make a statement that e.g. it has
> > > 4293 characters of textual content in it and you say it has only
> > > 1882 characters of textual content in it, then my assertion bears
> > > an authority that yours does not. True, it may very well be that you
> > > are correct and I might be incorrect, but that doesn't in any way
> > > change the fact that assertions about a resource made by the 
> > > owner of the resource bear special authority over assertions made
> > > by third parties.
> > 
> > Huh?  How?  Why?  What good does this do?
> 
> Well, to take one simple practical example, perhaps a manufacturer of
> a product has published some information about that product, and some
> trade publication has also published similar information about that
> product. Knowing who has said what may be very important in choosing
> what information one wishes to trust over the other. It may be that
> one trusts the trade publication more than the manufacturer. Or one
> may be aware of common errors made by that publication and wishes to
> let knowledge from the manufacturer take precidence in the case of
> a disagreement.

You appear to have changed from a view that the ``owner'' is always right,
to one where the ``owner'' may be wrong.  This is good, in my view, because
there are many cases where the ``owner'' has been wrong.  

> > > > Ironically, your claiming that there is such an authority is 
> > > > an example 
> > > > of a local ascription of meaning based on idiosyncratic use ;-)
> > > 
> > > I don't see that. I think you should re-read my post. Perhaps you
> > > missed what I was saying...
> > 
> > I don't think so.  I think that Mile's comments are exactly 
> > correct.  
> 
> Then it appears that you're both wrong ;-)

Oh, but the two sides are not equal here.  For you to win, there has to be
no effective dissent.  For Miles and I to win, all there has to be is a
multiplicity of opinions.  

> > All
> > that there is in your post is one person's feelings as to how URI
> > references are to denote.  Any other person could have a 
> > different view of
> > how URI references are to denote, and that other view would 
> > have just as
> > much validity as your view.
> 
> Sorry, but I believe that my view is born out by the specs, as outlined
> above. I also belive that my view is the most predominant view in the
> real world, both for developers as well as common users.

What specs?  The RDF Core Working Group voted, with no objection or
abstention, to remove that section of the docment.   Your view may be the
most predominant view of developers and users, but that doesn't mean that
it works in all cases.

> I also challenge both you and Miles to point to any normative content
> in any relevant spec which suggests that individuals can mint URIs
> using naming authorities they have no ownership or rights to, or that
> the denotation of a URI specified by the creator of a URI can be 
> questioned by any other party. Good luck.

I believe that such content does not exist.  However, I don't see what that
has to do with the issue at all.  First, I don't know of any normative
content of any relevant spec that mentions ``the denotation of a URI
[reference] specified by the creator of a URI [reference]''.  (In fact I
don't know of any spec that mentions the ``creator of a URI [reference]''.)
Second, Miles' and my view doesn't need any such support.  We are not
arguing for a restriction on how URI references get their denotation, which
would need some normative support, but instead are arguing against your
proposed restriction.

> > For example, here is a view that I happen to subscribe to
> > 
> > 	In any formal system, such as RDF, the denotation of a name
> > 	(including URI references as a special case) is left 
> > unspecified.
> 
> *In* RDF, yes, but not in the larger context of where and how that
> formal system is used.

What is this larger context?  Where is it defined?
(If you are arguing for some sense of ``intended meaning'', then, sure,
some notion of intended meaning is useful.)

> > 	Statements in the formal system, including RDF statements, serve
> > 	only as constraints on that denotation.  
> 
> But not the *only* constraints on that denotation. It is presumed (in
> fact IMO required) that there be system-external constraints on the
> denotations of the names (URIs) used as atomic elements of that system.

Again, where does this come from and how does it impact RDF?

> RDF does not happen "in a vacuum". 

Oh certainly, but neither, in my opinion, does RDF *have* to work (solely)
in the ``atmosphere'' that you are promoting.

> > Any agent (including
> > 	people) choose to believe certain statements, and thus 
> > every agent
> > 	can potentially have a different view of the denotation of any
> > 	particular name.
> 
> They *may*. But differing views in the *denotation* of URIs represents
> a breakdown in the system. RDF presumes (even if it can't garuntee)
> that URIs have globaly consistent, unambiguous, and immutable denotation.

Here is were we differ.  I believe that forcing a too-rigid view of the
denotation of URI references will cause a breakdown in the system.
Further, I don't believe that *RDF* makes any presumption on the denotation
of URI references.   

> RDF is *not* like natural language. RDF is intended to be much more
> precise, as are all formal languages, while still tolerating a certain
> degree of noise. Ambiguity and conflicting statements, even if tolerable
> to a certain degree, are nonetheless undesirable and detrimental to the
> optimal performance of that system, and therefore should be minimized
> as much as possible.

Well my view is that ambiguity and conflicting statements are unavoidable,
and, moreover, necessary to the functioning of the Semantic Web, and,
therefore, any regulations to minimize them should be avoided as much as
possible. 

> > For contrast here is another view, that might be closer to your view
> > 
> > 	There is some ultimate, timeless reality.  The 
> > denotation of every
> > 	possible name is fixed and unchangeable in this reality.  
> 
> Or rather, 
> 
>    There is a formal system for which the names in that system are
>    presumed to have unambiguous, consistent, immutable interpretation.
> 
> I don't see how the semantic web can offer any utility if it doesn't
> correspond to such a formal system.

Well, I believe that if the Semantic Web does correspond to such a formal
system then its utility will be severely limited.

> > This second view appears to be very nice, but there are quite 
> > a number of
> > problems in it, including how to provide a denotation of
> > 	http://www.whitehouse.gov/#46
> > and
> > 	http://www.dccomics.com/#clarkkent
> 
> I see no problems with the denotation of these URIs. They denote precisely
> what the creators of those URIs say they denote. If you don't want to 
> talk about the particular resources the owners say they denote, then don't
> use those URIs. It's as simple as that. You have *no* authority or basis
> to disagree with the denotation specified by the creator of a URI. Tough.

Well, then I guess I won't play in your semantic web.  However, don't
expect me to stop arguing for different views of the Semantic Web.

> > as well as how to deal with agents whose connection to this ultimate
> > reality is incomplete or incorrect.
> 
> This is yet another issue, disjunct with the issue being discussed. Feel
> free to start another thread to explore it.

Again, I view this as part of the same issue.  In particular, what happens
if the agent behind www.whitehouse.gov has an incomplete or incorrect
connection to reality?  Your view would cut off communication between this
agent and other agents that have a better connection to reality.

> Patrick
> 
> --
> Patrick Stickler, Nokia/Finland, (+358 40) 801 9690, patrick.stickler@nokia.com

Peter F. Patel-Schneider
Bell Labs Research
Lucent Technologies
Received on Wednesday, 2 April 2003 08:42:52 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:51:58 GMT