Key questions on Common Upper Ontology
4/20/2004 7:20
AM
Note from Paul,
We have posted the very constructive discussions at:
Of course, this is a very constructive discussion.
Several things are clear, but the interpretation of this discussion is subject to viewpoint.
From my viewpoint, the first point is leveraged by John Sowa’s comments regarding the observation that a great deal of the actual work being done with ontology is for the purpose of allowing structural interoperability between programs and their content. It is somewhat standard to refer to this as semantics – but this is where objections are raised regarding what can be implied about the use of the phrase “ontology about semantics” and “structural ontology”.
I would say that biological taxonomy is a structural ontology describing a theory about the organization of biological systems. Likewise other structural ontologies might allow “semantic” web services to know that the meaning of the sting “3.56668” is “this is a number”. The problem arises if we refer to this as semantic ontology BECAUSE we then have no words left to talk about the issue in linguistics related to the interpretation of experience while reading text, or while working in the garden. We have no way to distinguish the formal system and the natural system with language we use to talk about a certain class of real and practical issues. This class is the class of issues related to how humans and human communities use computers to perform tasks that are not defined by the computers but by other things, such as war and peace. { + }
For example, Leo said:
Describing this semantics, i.e. what is sometimes called the intended meaning of vocabulary terms, is exactly the job ontologies do for the Semantic Web.
And
this is precisely the point that is being made.
One
way to differentiate what we mean by ontology is to actually use the adjective “structural”
in all cases where we mean information that is useful when computers exchange
data structures about format and programmatic context.
For
example, the expression “this number whose value is 3.56668 is to be used as a value
in the function defined in the data structure called q.” This is ontology about the structure of
computer programs and content in computer programs.
Ontology
about “natural” semantics is then left free of the engineering doctrines that
attempt to formalize what many scholars observe to be not completely
formalizable. This is the issue touched
on by several, including Mike, [17]
Mike
said:
Since the introduction of pragmatics is so essential to the effectiveness
of a specific ontology, there will never be just one upper-level ontology
Mike
can mean several things here, without being in anyway unclear. The interpretation depends on
viewpoint. If the viewpoint is that no “ontological”
differences exist between the functions of a computer and the functions of natural
systems then many scholars of the natural sciences will object and suggest that
he is hopelessly confused. Of course, Mike
is not confused. But the issue remains
to be discussed.
One
interpretation of function: The functions of a computer
are to make state transitions within a finite set of states (physical states
engineered to have a one to one correspondence with the abstract set of all states
of a specific computer). The computer
is DESIGNED, or engineered to make this state transition by specific classes of
manufacturing processes and by programming languages.
A
second interpretation of function: The functions of any natural
system, whether alive or not does not matter, are determined by universal laws
such as gravity and structural constraints such as those we attempt to model
with string theory. Biological systems
have additional constraints and causes that science does not fully understand. The development of descriptive theory
regarding the function of biological systems, or natural systems, can thus give
rise to structural ontology that helps convey theories about these natural
systems.
There
is a fine line, however.
One
becomes hopelessly confused if one steps over the line and acts as if such a
structural ontology conveys “meaning” in a situated fashion, such as in the
meaning of an experience by a human of private knowledge. In linguistics, the issues are thus
separated into syntax, semantics, and pragmatics. The argument has been about confusions that occur if the notion of
semantics is castrated from the status of being “situated” and
experienced.
What
is missing in the Semantic Web community is the use of language that policy
makers will not be hopelessly confused when hearing.
My
opinion is that the careful use of proper language must explicitly recognize
that the natural world is not engineered, except in the construction of
specific machines.
Is
the “semantic web” concept of Tim Berners Lee a machine? I do not think that this is how TBL sees
this. There are two sides to this thing
that does not yet exist. The first side
is the human side, with all of the uncertainty, imprecision and natural
complexity of humans and human cultures.
The second side is a system of interoperable computational systems that
change state according to engineered and programmed finite state machines.
These
are “ontologically” quite different; and if we fail to continually recognize
this difference in our technical language; then confusion does arise. This is simply unnecessary and
counter-productive for everyone. Yes?
Paul
Prueitt
Director
BCNGroup.org
Chantilly
Virginia