Re: SUO: OpenCyc as SUO starter document
On 3/1/02 08:00, "Leo Obrst" <lobrst@mitre.org> wrote:
>
> I welcome the OpenCyc candidacy. A number of us have previously used the
> Cyc upper (perhaps pared of some stuff), and though it suffers from
> monolithicity (a general complaint, but at least at some level
> microtheories seem to be an attempt to address that issue) and perhaps
> inconsistency, it should stir the pot of issues SUO needs to address.
>
> Too bad we don't yet have generally available OntoClean-like tools which
> could slurp up ontologies and spit out indigestible (non-nutritional?)
> seeds: don't want to push this metaphor too far. Re: OntoClean, I do
> know that some projects are using the principles, per Guarino & Welty's
> CACM article, including OntologyWorks.
>
> I am still interested in modularity and the "lattice of theories" idea
> (and perhaps IFF will help in that regard), but I think we should
> welcome OpenCyc as a SUO candidate.
Ditto to Leo, John, and ChrisW. I haven't looked at the contents of Cyc in
a while, but have been anxious to look at OpenCyc. We (Ontology Works)
started out using Cyc in our original work but stopped - not because of
content (which is rich and good and largely well developed) but because of
computational reasons. I would welcome the chance to see if it would be
feasible to use it again as we (Ontology Works) aren't in the business of
making middle-level ontologies. A worthy project would be to attempt to
embed OpenCyc into the OntoClean framework (btw our tools could be used to
do that effectively).
With respect to monolithicity, could some of those (Leo, John) explain
exactly what you mean by it? The OntoClean stuff of Guarino & Welty, for
example, can certainly be considered monolithic in some sense. Now, Doug
McDavid told me it is not these kind of distinctions he's worried about when
he complained about "monolithic" approaches but rather middle-level entities
like "Organization". That makes sense to me - I think it will be hard to
legislate such things, unless there is a context mechanism built-in as Leo
suggests.
.bill