Re: An intuition pump

From: Jim Hendler <hendler@cs.umd.edu>
Subject: Re: An intuition pump
Date: Tue, 23 Sep 2003 12:21:41 -0400

> At 10:20 AM -0400 9/23/03, Peter F. Patel-Schneider wrote:
> >From: Jim Hendler <hendler@cs.umd.edu>
> >Subject: Re: An intuition pump
> >Date: Tue, 23 Sep 2003 08:48:05 -0400
> >
> >[...]
> >
> >>  Anyway, my problem is that given my simple world view, I cannot find
> >>  any interesting examples where Tim's solution would make smart people
> >>  like Bijan and Peter so upset, yet it clearly does, which is why I
> >>  ask for examples that can help a simpleton like me understand what
> >>  the pragmatic effects are
> >
> >[...]
> >
> >What makes me so upset with (a strong reading of) Tim's solution is that
> >eliminates many fruitful kinds of disagreement.  To communicate, one has to
> >use common vocabulary, but Tim's solution requires that the meanings of
> >just about all vocabulary terms are determined, in advance, by their
> >owners.
> 
> This is what confuses me - as I read it, the meaning isn't determined 
> with no disagreement - just the "referent" ...

I don't see how to determine the referent without determining the meaning.
In fact, I don't see how to determine the referent of most things.  All
that I know how to do is to determine some collection of information about
a term that I can then combine with other information I have.  (Well,
actually, I don't even really know how to do this, but I think that I can
do a better job of it than determining a referent.)

> >To pick on a similar example to Jim's, consider the vocabulary term ``Peter
> >Frederick Patel-Schneider''.  (This is actually a very useful vocabulary
> >term as it is almost certainly the case that there is only one person in
> >the world with that name.  Further, I had to go through an unusual, and
> >probably precedent-setting, process to assert my right to have that name.
> >There is thus a good case to consider this and related vocabulary terms as
> >being owned by me.)  Under (this strong reading of) Tim's solution, the
> >mere mention of ``Peter Frederick Patel-Schneider'', or any of its
> >variants, commits an agent to my view of the term.  (See my home page for
> >some of the consequences.  You will probably have to view the source of the
> >document as a browser is otherwise unlikely to give you the full impact.)
> >
> 
> 
> ... see - this is the thing - the term "Peter Frederick 
> Patel-Schneider" is some human construct and has nothing to do with 
> this argument as far as I can tell.  Your home page, 
> http://.../user/pfps/, is what we are talking about -- it is a 
> particular place in web space that I can find.  On your home page I 
> find:
> 
> 
> ...
> <foaf:Person rdf:about="#Peter_Frederick_Patel-Schneider">
>    <rdf:type rdf:resource="#PerfectBeing" />
> ...
> </foaf:Person>
> 
> referring to a class defined elsewhere on your page.
> 
> I think I should now be able to unequivocally state that THIS PAGE 
> claims that the person 
> http://.../user/pfps#Peter_Frederick_Patel_Schneider is an object in 
> the class http://.../user/pfps#PerfectBeing
> 
> Now, let's assume for the sake of the argument that I wished to 
> dispute your divinity for some reason.  I have an infinite number of 
> places on the web where I can state something pointing to this - for 
> example, my page could say
> 
> http://.../user/pfps#PerfectBeing owl:sameAs http://.../user/hendler#Moose
> 
> where I define Moose in whatever way I wish.  Therefore...

Sure, this would be fine. 

> >Under (this strong reading of) Tim's solution there is no possibility of
> >divergence of opinion concerning anything about a vocabulary term.  Any
> >agent who dares to disagree will just be inconsitent.
> 
> ... it seems to me we have a healthy disagreement expressed neatly in 
> web space without violating Tim's solution (as I understand it) 

I disagree.  Under a strong reading of Tim's solution your use of
http://.../user/pfps#PerfectBeing commits you to my meaning for the term.
You can (maybe) postulate additional information about
http://.../user/pfps#PerfectBeing but there is no way for you to opt out of
my meaning for http://.../user/pfps#PerfectBeing.

> because the things where we are referring to the same thing are 
> pointing to the "boxes" but when we disagree on content we're doing 
> it in different places - so an agent should be able to conclude that 
> .../hendler and .../pfps disagree on something relating to 
> .../pfps#Peter_Frederick_Patel_Schneider
> and I can tell from the various URI and dereferencing rules what that 
> disagreement is if I feel compelled to do so (note - many 
> applications may simply choose to ignore the disagreement and just go 
> on)

There is another possible reading of Tim's solution.  Under this very weak
reading, Tim's solution reduces to a commitment that absolute URI
references that appear in the Semantic Web are (absolute) names.  I agree
with this reading, of course, but I hadn't seen it proposed as the
preferred reading.

> >I am not against the deliberate self-imposition of a fixed common meaning
> >for vocabulary terms.  Even though this is not common in human discourse,
> >there are many cases where a fixed common meaning is useful, in particular
> >when systems with very limited reasoning power are employed.  However, I am
> >against the simple use of a vocabulary term committing one to a fixed
> >common meaning, and much in favour of an explicit mechanism (e.g., imports)
> >for this commitment.
> 
> this is what I cannot really understand -- it's where I'm really 
> looking for a use case that shows a difference --If you and I both 
> refer (in different places) to http://.../user/pfps#PerfectBeing we 
> don't need to import the page to see if we are referring to the same 
> thing.  

I think that we very much do need to determine what information we believe
about http://.../user/pfps#PerfectBeing before we can determine whether we
are using it to refer to compatible notions.

> On the other hand, if we then want to find all things that 
> are members of that class or what the implications of what we think 
> we are agreeing on, we certainly would need to do some other work, 
> which may well include importing
> 
> to put it another way - I love the fact that on the web we can, 
> unlike in NLP, know if we are referring to the same "box", and the 
> issues of when we need to look inside and what is meant where don't 
> seem to me to be drastically different on the SW than on the WWW. 
> Same URI refers to same "box", owner of URI is only one who can write 
> inside that box, anyone can look inside (dereference) that box to see 
> what the owner intended, any other box dereferenced that points to 
> that box is bound by same rules.

When are two systems in the Semantic Web referring to exactly the same box?
I see very, very few chances for this to happen.

> So, in a sense, I interpret Tim's approach as "The owner of the URI 
> gets to define what HE/SHE/IT means by that URI"   anyone else is 
> welcome to say things about the owner, the predicate, the URI, etc - 
> but they cannot change the "meaning" of that specific URI unless they 
> do it in their space - in which case it is their claim about the 
> meaning, not the owner's (and their URI is where they state what they 
> claim the meaning is).

I really don't understand what you could mean by ``meaning'' here.  I don't
see that it can be ``denotation'' or ``defining information'' or anything
else that I can relate to.

> Note also that a URI with nothing there (i.e. no dereferencerable 
> document or a non-document URI) works in the above anyway - the owner 
> makes no claim as to the meaning of the URI, but other systems 
> grounding at that URI can at least agree they are refering to the 
> same "box" despite it's null content

Well, I would like to be able to have the Semantic Web concern itself with
something besides boxes.  Perhaps all that you mean here is a common
vocabulary. 

> >Peter F. Patel-Schneider
> >http://www-db.research.bell-labs.com/user/pfps
> 
> 
>   -JH
> p.s Please note that I'm not responding to the issue of predicate vs. 
> subject/object as determining meaning - that one seems like it could 
> have some sort of import, but I haven't figured it out yet.

I see this distinction as a complete red herring.

peter

Received on Tuesday, 23 September 2003 12:45:07 UTC