W3C home > Mailing lists > Public > www-tag@w3.org > July 2003

Re: New issue - Meaning of URIs in RDF documents

From: pat hayes <phayes@ihmc.us>
Date: Mon, 21 Jul 2003 16:34:33 -0500
Message-Id: <p06001a02bb420606b595@[10.0.100.23]>
To: "Patrick Stickler" <patrick.stickler@nokia.com>
Cc: "ext Tim Berners-Lee" <timbl@w3.org>, <www-tag@w3.org>
>
>
>
>This depends on what you mean by "contexts". If you mean that I can 
>send one person an email saying (in RDF)
><http://example.com/foo.rdf#bar> pantone:color "blue426" .
>and it can mean one thing and I send it to another person it can 
>mean something else,
>then we do not have system of communication which has any properties at all.
>
>
>
>Exactly.
>
>Pat, we've had this discussion at length several times before.

I know, Patrick.

>Your assertion that the RDF MT does not presume a single interpretation
>is quite correct. Any such constraint would make RDF useless to a
>wide number of applications which are not part of the semantic web. And
>even though RDF is a foundational component of the semantic web,
>its use is certainly not limited only to the semantic web.
>
>It is the SW Architecture that asserts the presumption of

(1)

>of a single shared
>interpretation,

(2)

>of a virtual, ever changing, global graph consisting of all
>statements accessible as part of that semantic web of knowledge.
>

These are not the same claim. (2) is the SW vision I agree with. (1) 
is just plain false. Agreeing to shared beliefs is not agreeing that 
these beliefs have only a single interpretation.

>All agents operating on knowledge which is "on the semantic web"
>are committing to presume and honor , as well as possible, a single
>interpretation of that  virtual, global graph of statements.
>

No, they are NOT.  Not only are they not doing that, it would be 
*impossible* for them to do that. Provably, mathematically, 
impossible: it would violate Goedels' two completeness theorems.

It kind of works for the Pantone example because Pantone color URIs 
*are* genuinely grounded in the physical world: you can buy a gadget 
which will check them for you (eg 
http://www.colormall.com/product1.asp?CustomerID=73550&ACBSessionID=fhPDkHUqUR439ii1q5UI&SID=1&Category_ID=51&Product_ID=444) 
but most URIs are not grounded like this. And even the Pantone 
example is doubtful, in fact, if you are need to get detailed, eg 
what if I have a color ontology which distinguishes the categories of 
color, solid color, RGB mix color and CMYK mix colors as distinct 
categories? *Now* what single resource does pantone:color "blue426" 
denote, exactly? What if it has to communicate with a physics 
ontology which says that color is a property of the reflectance 
spectrum of a surface? How about a psychophysics ontology that 
defines color as a behavioral response to a class of visual stimuli? 
Or an artists' ontology that classifies colors as mixtures of named 
pigment types? Or an art glass ontology which distinguishes 
transparent colors from surface colors?

>This is one benefit of having a standardized mechanism for accessing
>statements which are "on the semantic web" such as URIQA as it
>provides an explicit mechanism by which that commitment is expressed.

What commitment? To a single interpretation?? How can that POSSIBLY 
be expressed?

>Any agent obtaining descriptions of resources using the URIQA extensions
>is explicitly agreeing, by that act, to presume a single 
>shared interpretation of
>all statements on the semantic web and honor the authoritative semantics
>ascribed to any resource as exposed in its authoritative URIQA accessible
>description.
>

No, it isn't. I don't even know what URIQA refers to, but I know its 
not doing what you say here, because nothing can do this, because 
this is impossible.

>Any knowledge producer exposing descriptions of resources using the
>URIQA extensions is explicitly agreeing, by that act, to presume a single
>shared interpretaion of all statements on the semantic web and honor the
>authoritative semantics ascribed to any resource as exposed in its
>authoritative URIQA  accessible description.

Honoring the semantics, sure. Agreeing to a single interpretation, never.

>There will be other uses of RDF, OWL, and related SW technologies which
>will not be part of the Semantic Web. But all agents, systems, models,
>operations, etc. which are part of the Semantic Web *must* presume
>a single shared interpretation of  all statements,  and a consistent,
>unambiguous  denotation for each URI.

Patrick, you don't know what you are talking about. Sorry, but 
there's no other way to put it.

>Does that mean that there won't be noise, errors, bugs, contradictions,
>etc on the semantic web? Of course not. And in fact, without this
>fundamental presumption of a single common interpretation, one
>is not even able to detect or assert the presence of any contradictions,
>since one can simply conclude that all statements are valid/true,
>based on any arbitrary number of interpretations.

Total baloney.  Detecting contradictions is done by checking 
entailments, which does not only not require the assumption of a 
single interpretation, but in fact couldnt be done if there was only 
a single interpretation, since it is in effect a process of searching 
through possible interpretations.

>
>Without the presumption of a single consistent interpretation for all
>semantic web knowledge, the semantic web is completely useless
>as to its intended purpose.
>

It would be useless - or more accurately, irrelevant - WITH this 
assumption, since if this assumption were correct then there would be 
no need to check anything for consistency or even to communicate 
anything: all agents would already know the truth of all truths and 
the falsity of all falsehoods, and there would be no need for 
communication or inference to take place at all. We would all  be 
omniscient: if there is only one possible interpretation, then we all 
know all the truths that can possibly be known. To learn anything new 
would be impossible, since one cannot constrain the interpretations 
even further than a singleton.

Pat

>Regards,
>
>Patrick
>
>--
>
>Patrick Stickler
>Nokia, Finland
><mailto:patrick.stickler@nokia.com>patrick.stickler@nokia.com
>


-- 
---------------------------------------------------------------------
IHMC	(850)434 8903 or (650)494 3973   home
40 South Alcaniz St.	(850)202 4416   office
Pensacola			(850)202 4440   fax
FL 32501			(850)291 0667    cell
phayes@ihmc.us       http://www.ihmc.us/users/phayes
Received on Monday, 21 July 2003 17:34:37 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 22:56:00 UTC