[Top] [All Lists]

Re: [ontolog-forum] Danger of URIs in mission-critical applications

To: "[ontolog-forum]" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: "John F. Sowa" <sowa@xxxxxxxxxxx>
Date: Thu, 09 Jul 2009 22:09:35 -0400
Message-id: <4A56A2DF.2010003@xxxxxxxxxxx>
Ron, Ed, and Pat,    (01)

JFS>> We still have to develop SURIs (Secure URIs).  But until they
 >> become available, we can reduce the security risks, improve
 >> efficiency, and shorten the names by grouping them in contexts.    (02)

RW> Isn't that how it is done now? Are we not just describing
 > namespaces?  The only tricky bit is knowing when to refresh the
 > cache. Most often, in my line of work, that is a manual decision
 > that is determined by my choice of time to have my world disrupted
 > by updated definitions.    (03)

Yes, but we have to accommodate a wide range of successful practices
and paradigms for using unique identifiers in various fields.    (04)

Many unique identifiers, such as 'H' or 'Na' for chemical elements
or 'Homo sapiens' for biological species, have been standards for
centuries.  For identification, those strings are the ideal URIs,
and any new technology that comes along should include them as part
of the foundation.  An address-like URI, as used on the WWW, should
point to a document (or namespace) that contains a few dozen or a
few thousand such identifiers.    (05)

If any address-like URI is used for 'H', it would be a URI for
some official document about chemical elements with 'H' as the
last part of the compound name.  Within that document or namespace,
'H' would be the local identifier.    (06)

EB> The URI is an identifier that "can" be dereferenced but may
 > be sufficient in its uniqueness that actual access to the defining
 > resource is very infrequent.  I need to know we are both referring
 > to the same "gram" but I don't need to retrieve the definition to
 > know what it is.    (07)

I agree.  Various communities and institutions have been developing
conventions for unique identifiers for many centuries.  Any system of
knowledge representation or ontology must support such systems, not
replace them.    (08)

EB> And oh yeah, this is such a wonder of modern technology that we
 > also have GUIDs and UUIDs and UPNs and URFIDs and DUNSIDs and a
 > dozen other unique machine-readable terms, each for someone's
 > favorite community and domain.    (09)

Such identifiers and conventions for supporting them have been
around for millennia.  They arose as part of the notations by
the Sumerians, Egyptians, and Chinese, and they will continue
to proliferate indefinitely.  Any system of identifiers must
accommodate them.    (010)

EB> In so many words, the problem with URIs is that they are not
 > *yet* supported by a technology that is widely adopted and
 > guarantees that they have value other than as a "guaranteed
 > to be unique" string of characters.  (John's points about
 > security and reliability and expectations enter into this.)
 > And there are many "guaranteed to be unique" string "technologies"
 > that are competing to be THE reference identifier for lots of
 > different things.  What URIs have right now is "promise", not
 > "value".    (011)

I agree.    (012)

This thread started with my response to Kevin Keck, who criticized
some ontology proposals for not using W3C-style URIs for each
term in the ontology.  I believe that the unit of definition
and update for ontologies is a complete theory or microtheory,
since no single definition can stand on its own -- each term
is defined by its relations to every other term in the theory.    (013)

A document that contains the entire theory may have an address-
like URI, but the individual terms should be simple strings
within the namespace of the theory.    (014)

For security, various conventions can be used to prevent or
at least to detect unauthorized modifications of the document
containing the theory.  But once the theory is loaded into
some local cache and verified as authentic, no further
checking would be necessary for individual terms.    (015)

PC> Among other problems, there can be any number of URIs that
 > are intended to mean the same thing, but there is no way to
 > determine whether or not they actually do mean the same thing.
 > What is missing is some kind of agreement on how to specify
 > meanings with sufficient detail to be unambiguous to people
 > and machines.    (016)

In science, the definitions of the most fundamental terms are
always evolving and changing as new discoveries come along.
The terms like 'mass', 'energy', and 'force' were around for
many centuries before Newton.  But his theories gave far more
precise meanings to those terms, which were rapidly accepted
by the majority of physicists.    (017)

But the new discoveries at the beginning of the 20th century,
kept the same words, but forced radically new definitions upon
them.  Today, physicists and engineers use those same terms
with a variety of different and *inconsistent* definitions
for different purposes.    (018)

Today, physicists know that Newton's equations are "false"
when pushed to certain extremes of motion or size.  But
they also know that for "midworld" applications (i.e., not
too large, too small, or too fast) the errors in them are
less than the errors of typical measurements.  Therefore,
they use Newton's equations instead of the "more precise"
equations of relativity and quantum mechanics.    (019)

That is the case in physics.  All other empirical subjects
are in a far more chaotic state than physics.    (020)

PC> That is a function that a common foundation ontology
 > could serve; the elements of an FO could be labeled by URIs
 > or any other unique ID system, but the meanings would have
 > to be derived from the agreed interpretation of the basic
 > defining elements of the FO.    (021)

How could anyone possibly agree on the details of an FO?
Physicists can't agree to common definitions, since
they knowingly prefer to use theories that they also
"know" are "false".  And when you go from physics to any
other field, the situation gets progressively more
complicated.    (022)

In every field of science and technology, the practitioners
have a large bag of theories.  They know that most of those
theories are mutually inconsistent.  But depending on the
problem at hand, they quite happily pick and choose one or
another from their stock of inconsistent theories.    (023)

If this is the state of the art in physics, how can you possibly
assume that any group of experts in any field could agree on
a fixed set of definitions for the terms in their field?    (024)

The best that can be done is to record and catalog all the
options and variations.  Let the experts in each field
provide guidelines for picking and choosing theories that
have proved to be useful for various problems.    (025)

John    (026)

Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (027)

<Prev in Thread] Current Thread [Next in Thread>