On Mon, Dec 10, 2012 at 11:48 AM, Sandro
Rama Fiorini
<srfiorini@xxxxxxxxxxxx>
wrote:
Hello
all!
There is a question we have been
discussing in our group for a long time,
and we would be glad to have your opinions
on it: Can ontologies really
contain individuals?
I think that this question requires alot of
precision, in order to usefully consider it.
First of
all, when somebody says
'really'
there is a suspicion that there is a
philosophical axe in the background. What
is the supposed difference between
Can ontologies REALLY contain individuals?
and
Can ontologies contain individuals?
?
I have always been bothered by that word,
'really'. I was once told 'positions are
not 'really' objects!' (because the
represent a relationships between an account
and a financial instrument.)
Second,
Ontologies
They 'contain' individuals of only a few
kinds: at least three kinds of individuals
1. Structuring Rules
2. The entries in the ontology that refer
to things that exist outside the ontology
(which may include Identifiers, names,
Definitions, etc. accordingto the
structuring rules.)
3. Assertions about Relationships between
entries
So,
I would phrase the question as:
can one of the categories of entry be an
entry for an individual,
And, I would say this is one of the
structuring rules for the ontology.
We know that many of the ontology
representation languages provide
constructs to represent individuals.
However, an ontology is a theory about a
conceptualization, which in turn could be
loosely seen as a structure of
generalizations about a given domain.
Generalizations
usually refer to concepts; not to
individuals.
I would
think that almost by definition, a
'generalization' is something that generalizes
from a set of individuals. So, I agree with
Alex. OTOH, most of the entries in an
ontology will be entries about classifiers
of some kind, not individuals.
Thus,
one might ask how a conceptualization can
have individuals that could be
specified.
I think
that individuals are generally **not**
'specified', but rather identified, often
with some kind of locatability. For
example, Frege, "On Sense and Reference".
This is the real problem: Individuals are
identified by a set of features that let you
know whether different experienced phenomina
are experiences of the 'same' individual or
a different one. For physical things,
these features may include spatio-temporal
contiguity. We do not usually require
continous movement through space time to
determine whether a thing is of a certain
'kind'.
Specifications can be tied to a temporal,
spacial location, like a description of the
Golden Gate Bridge, or, if the individual is
abstract, such as the poem 'My Last
Duchess'. the specification, by definition,
has only one instance, and is not even
distict from its one instance.
On the
other hand, we also acknowledge the fact
that including individuals in ontologies
might be necessary in some cases, such as
when a concept is defined in relation to
specific particular (e.g. "Former US
President", "Red Car").
Furthermore,
there are some authors actually
questioning this sharp distinction between
universals and particulars (there some
papers about this in FOIS 2010, I guess).
I think
starting with Aristotle, Categories,
through say Davidson Essays on Actions and
Events, and many others who deal with
individuals and concepts both, and try to
show how they are related and continuous,
and Hillary Putnam, 'Ethics Without
Ontology' as well as those who focus one of
the two,for example, Quine, "On what there
Is" on the side of individual things, along
with all 'nominalists' to Plato and the
OMG, who seem to think individuals are
second class citizens, and other 'realists'
who treat the concepts as the really real.
OTOH, many on
this forum have pointed out that all this
philosophical 'debate' should be avoided.
However, I personally think that there is an
implicit, entirely pragmatic, ordinary
language ontology, sort of
anti-philosophical ontology built into our
practice. And, in that world, there are
obviously important individuals, and
characterizations of types of
individuals, and clubs that contain of
individuals, which are themselves
individuals, etc. And, since all of these
things exist, they are themselves
individuals, for example, each individual
type, which has an identity as an
individual, as well as applying to many
other individuals. The practical questions
are things like: how to identify
individuals, apart from a characterization
of a type that just happens to have only one
instance.