Back to Jim
Schoening statement
Congressional testimony from
DARPA Director, Tony Tether ->
.
Jim (Schoening),
It was good to see you again and hear your talk (though it had to be somewhat rushed). Here are my answers:
To your
first question:
1.
Would
a 'good enough' common upper ontology provide benefits? Or rephrased, is
there any reason why we would not want the Army, DoD, or Federal government to
adopt a common upper ontology (for systems needing data/semantic
interoperability ) .
Yes,
I think it is a very important step. If it becomes widely used, one of
two things will happen: (1) It will be adequate for most people and
they will find it to their advantage to use it, for the obvious reasons that
cause people to use standards, or (2) It will be changed in some ways in the
future, for the obvious reasons that cause standards to be changed.
I think that (2) is less likely than (1) - at least for some time - because high level ontologies are less likely to succumb to quick changes in technology or accepted terminology than mechanical or digital devices; but even if (2) occurs, it should be possible to change the ontologies that conforms to the old standard by a single program, run once on the existing ontologies. A conformance test can also be easily produced to detect whether the old or new standard is being used.
To your second question
2. Can we achieve cross-domain
semantic interoperability without a common upper ontology? (Note:
Any two ontologies can be mapped, but this becomes an n-squared problem, which
can't scale up.)
It
is possible to make the mappings, as you state; but it is far better to have a
standard. Suppose that the Army, Navy, and Air Force each has its own
ontology because the Army has used Boeing, the Navy has used Northrop Grumman,
and the Air Force has used Lockheed-Martin as ontology contractor.
Then
there will always be disputes about which one should be the used to create a
merged ontology or whether patches should be used to achieve interoperability,
when clearly they need to interoperate.
It
is the same problem that confronts the "stovepipes'" which have
always plagued DoD and the armed services (and almost any big organization) and
have gummed up common data base systems (including things as fundamentally
similar as payroll systems, which have been pared down a bit over the years,
but not without expense and not entirely) and caused unnecessarily redundant
computer programs to be produced at taxpayer expense.
Heaven
knows that there will be enough differences in the lower details of the
ontologies anyway (as there are always, even between me and my wife) but having
a standard upper level will help, and - one hopes - the upper level will
proceed downward with time.
As
an additional "plus", I think the whole standards effort can help in
the overall ontology development effort, that is, the idea that it saves money
and grief in the future if an ontology of a large system is developed early, at
the requirements stage and made explicit at that time (even if it has to be
modified as an understanding of the system changes with the project), whether
in military or civilian government or commerce.
Exemplary
ontologies utilizing the Common Upper Ontology should be produced in some
standard notations to promote this outcome.
Larry H. Reeker, Ph.D.
Senior Computer Scientist
Office of the Director,
Information Technology Lab
National Institute of Standards & Technology
E-mail: larry.reeker@nist.gov