On 31 Jan, at 10:44 , Pat Hayes wrote:
>> Hi Everyone,
>>
>> ... I do think, though, that some
>> measure of correction of logical constructions is probably also
> necessary,
>
> Amen to that. But it is very hard to see how this is to be done. I
> REALLY wish there were a nontrivial and useful notion of how to
> measure 'correctness' of an ontology. It is not enough to just say,
> it is correct if it "fits the facts" in some sense, since ontologies
> may be based on very different, possibly mutually contradictory,
> conceptualizations, and yet both fit the facts perfectly well. (01)
Yes, exactly. And even the idea that there are theory/ontology-
independent "facts" relative to which an ontology can be deemed
correct is a HIGHLY dubious notion. Granted, there are arguably
facts that are largely ontology-free, or at least assume a rather
uncontroversial "commonsense" ontology that we all share in common in
virtue of being wired in roughly the same ways and speaking a variety
of languages that seem pretty unproblematically intertranslatable --
mundane facts about trees, persons, cabbages, kings, etc. But the
kinds of facts that can be used to confirm or disconfirm an ontology
typically *presuppose* large portions of some ontology, typically the
very one in question. In the hard sciences, for example, facts about
meter readings, cloud chamber photographs, what is seen via a
microscope (let alone an electron microscope) already presuppose a
huge amount ontological baggage about how the various apparatus work,
what exactly is being measured, and so forth. So it is not clear
that any coherent, objective notion of correctness is possible that
doesn't already beg a lot of questions. (02)
Note that I don't at all think that his means that everything is up
for grabs, or that there is no such thing as a correct, etc. I just
don't think there is any ontology-independent way of *establishing*
it -- it's just not possible, once again, to express the facts an
ontology is supposed to fit to be considered "correct" without
already assuming some ontological framework. And even if we agree on
a minimal ontology O for expressing a certain base of facts, there
might, as Pat notes, be ontologies O1 and O2 that extend O in
incompatible ways but which fit the fact base equally well. (As Pat
is well aware, both of these points -- the ontology-dependence of
"facts" and the possibility of empirically indistinguishable but
logically incompatible ontologies -- are well known and extensively
discussed in the philosophy of science, and trace back at least to
Duhem in the 19th century. The latter point in particular was
resurrected most notably by the logician and philosopher W.V.O. Quine
in the late 20th century.) (03)
I think there are several more useful notions than correctness that
might be more productively developed. More or less of the top of my
head: (04)
* Semantic clarity: Can the proposed concepts of an informal
ontology be fleshed out rigorously in such a way that information
expressed in terms of those concepts can be objectively shared and
reasoned upon? (05)
* Logical coherence: Are the various concepts of an ontology (when
rigorously spelled out) consistent, both individually and jointly? (06)
* Empirical adequacy: Relative to some assumed collection of facts
-- and therefore relative to some assumed underlying O -- is a given
extension O' of O compatible with that collection of facts? (07)
* Practical applicability: Does the ontology serve its intended
purpose? (08)
There are surely others, and these could use a lot of tightening up,
but perhaps they are a reasonable first cut at a list of useful
evaluative criteria for ontologies. (09)
Chris Menzel (010)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Subscribe/Config: 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 Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx (011)
|