[Top] [All Lists]

Re: [ontolog-forum] Foundation ontology, CYC, and Mapping

To: "[ontolog-forum]" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: "John F. Sowa" <sowa@xxxxxxxxxxx>
Date: Sun, 31 Jan 2010 17:32:29 -0500
Message-id: <4B6604FD.5000507@xxxxxxxxxxx>
Pat,    (01)

Let me start with the last two sentences of your note:    (02)

PC> Anecdotal stories of someone trying something and failing are
 > quite useless as evidence; the tell us nothing about what works
 > and what doesn't, and why.  They are, of course, amusing.    (03)

They are far more than anecdotal.  Basic principle:  low-level
ontologies are useful, but detailed axioms at the upper levels
are of minimal value.    (04)

Look at the $70 million invested in Cyc from 1984 to 2004 with
no deployed applications.  After a review of that work in 2004,
the gov't agencies made a sharp cutback in funding, and the
Cyclers have since turned their attention to doing what they
should have been doing for years:  focusing on domain-dependent
ontologies for specific applications.    (05)

Furthermore, we have many theoretical analyses about the nature of
language and logic, which show very clearly where the problems lie.
 From Peirce ("Symbols grow") to Wittgenstein (language games) to
Terry Winograd (his 1972 book _Natural Language Understanding_
to an anti-AI approach in his publications since 1986) to the
linguist Alan Cruse (microsenses) to the formal semantics pioneer
Hans Kamp (http://www.illc.uva.nl/lia/farewell_kamp.html ).    (06)

Another case in point is the work by Bill Andersen at Ontology
Works.  Bill and his colleagues had worked with Cyc at one of
the gov't agencies, and they had focused on ways of developing
applications (along the lines that Lenat should have done).
They formed their own company, and they are now quite successful.
But Bill said that their original attempts to develop a general
upper-level ontology were not useful.  They are now much more
successful in working on domain-dependent ontologies for
applications.    (07)

With that kind of evidence, no funding agency is going to cough up
$30 million for a study project whose chance of success is somewhere
close to 0%.    (08)

PC> For the near future, I feel certain that a serious effort to
 > both logically specify and linguistically define the FO terms
 > precisely will result in few misinterpretations, and allow the
 > stable meanings that are required for accurate general semantic
 > interoperability.    (09)

Your feelings aren't even based on anecdotes.  You keep using the
term "stable meanings."  But engineering is devoted to innovation.
By definition, their goals are antithetical to stable meanings.
In computer engineering, meanings change with every release of
every component.  Just one more example:    (010)

At VivoMind, we develop our software to run on Unix-like systems
(mostly Linux and Mac OS X) and we port it to Windows.  We use
the "same" compiler (GCC) for the C code.  But the GCC libraries
for the Mac have been tweaked to make them more Windows-like
than Linux-like.  If you're just compiling one little program,
you won't notice the difference, but if you're compiling a large
system, you run into these hassles every time.    (011)

We've had ISO standards for C for many years, but all the compilers
have glitches that keep them from being completely compatible.
Even when you're using the same (GCC) family of compilers, you run
into glitches with different releases on different systems.    (012)

This is, of course, system programming.  But the same kinds of
subtle mismatches occur in application programming.  If you think
a Foundation Ontology based on 2,000 English words will magically
solve such problems, you're out of your ever-lovin' gourd.    (013)

John    (014)

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    (015)

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