I think it would make sense to set up a structure as you describe and
populate it with such terms as can be defined semantically from
authoritative industry sources i.e. industry standards, along with
annotation of the provenance of those semantics. This would then also
become a valuable tool for those same industry bodies and standards
groups, I think. A vital component of this would be change management,
such that when the competent authority makes a change or an addition to
their semantics, this can be picked up and propagated through the
resource and any developments that have made use of this resource. A
tall order perhaps, but not as tall as maintaining an isolated huge
For the most part such terms would also be more relevant to how
information is passed between computers, than something from the broader
and fuzzier world of the human language dictionary, I would venture to
suggest. They would certainly be simpler. (03)
I like the idea of including case study information on interoperability. (04)
John F. Sowa wrote:
> Pat C, Ali, Dave, Ed, Rob, and Pat H,
> Before getting into the details of what I've been arguing against, I'll
> summarize what I've been arguing for in many notes over the years:
> 1. A loosely axiomatized hierarchy of categories (predicates,
> relations, functions, types, or whatever they're called in the
> version of logic that is being used). Those categories may be
> associated with lexical resources such as WordNet, Longman's
> "primitives", Anna Wierzbicka's "primitives", Roget's Thesaurus,
> or many other resources that have been or will be developed.
> 2. For any particular application, more detailed axioms are needed
> to support precise reasoning. Those axioms are organized in
> small, modular theories (or 'microtheories' in Cyc's sense).
> Those theories can related in a finite hierarchy (of which
> the infinite lattice is the theoretical extension).
> 3. All the resources in #1 and #2 are stored in a repository,
> with all the documentation, case studies, testimonials for
> and against, stored with them. All the relations that Ali
> proposed for COLORE and any others that anybody else might
> find useful could also be stored with them.
> 4. Tools for using, analyzing, relating, combining, testing,
> extending, and editing the term hierarchy and the hierarchy
> of theories should also be developed and made available.
> 5. Further additions, extensions, and modifications that anyone
> might invent or implement may be added as time goes by.
> I believe that this kind of system could serve the requirements
> that Pat C. would like to achieve, it can accommodate formalized
> versions of theories based on Longman's list, and anybody who
> finds them useful could use them and write a testimonial for or
> against them along the lines of point #3 above. But it doesn't
> require anyone to limit their choices to any predetermined set
> of terms (whether primitive or not).
> For interoperability, anyone who has stories about successes
> or failures can document them in the repository, and further
> decisions can be based on the accumulated experience.
> DMc> We've taken a similar, but less ambitious, approach to Pat
> > and created what we call a minimalist upper ontology, but perhaps
> > Foundational Ontology would be a better name. We started with
> > some of the same inspirations (Wierzbicka and in our case the
> > observation that Tok Pisin the official language of Papua
> > New Guinea, where I spent a lot of time, has only about 2000
> > words was similarly encouraging)
> That is useful information. But as the following discussion shows,
> what you're doing is closer to point #1 above than to a formally
> defined Foundation Ontology as Pat described it.
> DMc> We've found a foundation ontology with about 100 properties and
> > 100 classes typically needs another 100 properties to cover the kind
> > of enterprise ontologies we've been working with, and several
> > hundred very specific categories to cover the kinds of distinctions
> > people generally create in commercial systems, that aren't easily
> > reduced to axioms.
> EB> Which, I assume means you didn't. So you have several hundred
> > very specific "primitive" classifications, i.e., "undefined" in
> > any formal sense?
> DMc> Yes, but this is down from several thousand, so it seems like progress.
> EB> I realize that by guessing at what Dave meant, I may be
> > constructing the strawman I am knocking down. My apologies if
> > I am well wide of the mark.
> DMc> It's not perfect, but I don't see why you wouldn't start with
> > an FO that has been rigorously debated and disambiguated as
> > a starting point. Our experience is that it was useful.
> What I have been arguing *for* is a loosely axiomatized upper
> level that is closer to what Dave has been doing than a very
> precise FO that has detailed axioms.
> I definitely support the option of detailed axioms, but only
> in the lower-level microtheories (point #2 above), not the
> in the looser hierarchy (point #1).
> RF> I would rather emphasize the similarities in our arguments,
> > viz. no single theory which explains all others.
> At that level of detail, I would agree.
> JFS>> Football requires far more than naive physics. Those qualitative
> >> versions of physics were proposed for understanding the verbal
> >> patterns in ordinary language. But a robot that could play
> >> professional football would need *real* physics at a precise,
> >> detailed level.
> PH> Playing football at an expert level requires specialized football
> > knowledge, not physics.
> I agree. By "real physics", I meant the kind of physics that we
> recognize almost instantly and respond to with our muscles, not
> with abstract concepts, propositions, and formal logic.
> PC> The estimate is taken as the aggregate of several studies
> > estimating interoperability losses in different industries...
> > http://www.bfrl.nist.gov/oae/publications/gcrs/04867.pdf
> > http://www.nist.gov/director/prog-ofc/report99-1.pdf
> I browsed through those documents, and I agree that they make
> some interesting points. But none of the issues they discuss
> are related, directly or indirectly, to your proposals for an FO.
> PH> People seem to simply take it as obvious that ontologies
> > create interoperability, but they do not. At the very least,
> > interoperability requires code to be written. Code, unlike
> > ontologies, actually does something. Where exactly does
> > the ontology become involved? Can you sketch a scenario?
> Those are excellent questions.
> PC> If you have a better estimate [of the savings],
> > please let us know.
> My estimate is a savings of $30 million by not funding
> your proposal.
> Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
> Config Subscr: 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 join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
> To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx
89 Worship Street
London EC2A 2BF
Tel: +44 (0) 20 7917 9522
Mob: +44 (0) 7721 420 730
Registered in England and Wales No. 2461068 (07)
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx (08)