On Fri, Jan 26, 2007 at 03:51:32PM -0500, Obrst, Leo wrote:
> I've changed the subject line of this message to better describe the
> topic.
>
> In general, I tend to agree with you both concerning profligacy of
> models -- but for logical theories. And it seems to me that ontologies
> add (or subtract) something to that notion, i.e., their concern with
> the real world *. And yes, I don't want to get into long philosophical
> arguments either, amateurish or otherwise. And I know that "real" begs
> the question. But I still think there is something crucial there.
>
> I think one needs to use the best ideas, principles, analyses that
> come out of formal ontology and formal semantics in the development of
> (little o) engineering ontologies. For example, I think the OntoClean
> methodology is an excellent step in the right direction, focusing on
> meta-properties that semantically distinguish choices in the
> subclass/is_a taxonomic backbone of an ontology. Similarly, the 3D/4D
> distinctions and analyses that come out of formal ontology are
> excellent for considering how you develop an ontology, as are notions
> of intension/extension, possible worlds, etc., distributed/cumulative
> readings, abstract individuals, fusion operators, interval semantics,
> etc., from formal semantics. And to me, all these are rooted in the
> real world. They are important precisely because they are tools that
> try to help our understanding of some real domain **.
>
> * Real world to my mind allows for conditionals, abstract realms,
> fictional or dream domains or other cognitive states (internal
> experience is still real experience). Mathematics to me is real, even
> though not all mathematics is known. In a very real sense, I think
> even mathematics is like science, partially empirically determined or
> discovered.
>
> ** One potential example: phlogiston theory. I think that it is or can
> be a logical theory, even today. But is it an ontology? Nowadays, I
> would say no, because it doesn't represent our best science of what is
> real. Was it an ontology at one time: yes, quite possibly. (01)
I am in general agreement with almost everything you say here, Leo, but
I simply think your additional criterion for distinguishing ontologies
from mere logical theories will remain so vague as to be useless and
will, furthermore, generate needless controversy. In this very post, in
fact, it seems to me that your otherwise cogent post above illustrates
the problem. You note first that ontologies add something to logical
theories, viz., "their concern with the real world." The weasel word
here is "concern". What is it for a theory to concern the real world?
Surely, on a relative natural understanding of "concern", phlogiston
theory is, even today, concerned with the real world. It provides an
explanation for certain empirical phenomena concerning combustion (among
others); it has simply been supplanted by a better theory that explains
more without appeal to ad hoc fixes. So "concern" must mean something
stronger. But what? If you want to say that our current theory of
oxidation count as ontologies, but phlogiston theory does not, then I
can only think of two things that distinguish them -- it is simply (as
noted) a *better* theory: it explains more. So perhaps the idea is that
a logical theory is an ontology if it is a better theory of its domain
than any other theory. But now the problem is that formal ontologies
typically play no *explanatory* role at all. What does, say, a wine
ontology explain? Explanation seems to me to be a role that only a few
"elite" ontologies play. (02)
So all I can imagine that you mean by "concerns the real world" is
simply "true". But surely that is *way* too strong a criterion; we have
no way to guarantee that even our best scientific theories are true --
it is reasonable to *believe* they are, but the history of science
suggests that even very well corroborated theories can turn out to be
false in important respects. Moreoever, when it comes to ontologies of
largely artifactual domains, like wine or automobiles, the whole issue
of truth just doesn't have a lot of purchase, as the concepts that
dominate in those domains are ones that *we* made up. The test of
quality for ontologies in these domains is not so much truth as the
depth and accuracy of the manner in which an ontology represents the
logical connections between the concepts in the given domain. (03)
So I'm with John here. A formal ontology is simply a logical theory;
*any* logical theory. This definition is completely clear and precise
and cuts through all the generally useless haggling that results
whenever anyone asks what an ontology is. Moreover, I think it is an
additional virtue that it cleanly separates what ontologies are from
their quality -- and that's where all the action is, after all.
Ontologies can be subject to a myriad of evaluative criteria. Some are
just silly (recall the Monty Python skit in which John Cleese (as Miss
Elk) seriously proposes the "theory" that all brontosauruses were rather
thin at one end, MUCH thicker in the middle, and then thin again at the
far end :-); some are logically inconsistent; some are stronger than
others (by notion or another of strength); some are finitely
axiomatizable; and so on. By trying to require that logical theories
meet some further intentional, or epistemological, or metaphysical
condition before they can count as fully-fledged ontologies seems to me
turn "ontology" into a needlessly honorific title that invites
controversy and unnecessary disputes -- who, finally, is the arbiter who
decides what candidate ontologies shall deserve the title? Much easier,
it seems to me, is to level the semantic playing field here -- let every
logical theory count as an ontology from the git go -- and let them all
compete. The best (read: most useful) will emerge as the survivors in
the ensuing process of natural selection. (04)
Chris Menzel (05)
_________________________________________________________________
Msg Archives: http://ontolog.cim3.net/forum/ontology-summit/
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontology-summit/
Unsubscribe: mailto:ontology-summit-leave@xxxxxxxxxxxxxxxx
Community Files: http://ontolog.cim3.net/file/work/OntologySummit2007/
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2007
Community Portal: http://ontolog.cim3.net/ (06)
|