Re: SUO: Re: Negotiation Instead of Legislation
Leo,
I have a very high regard for the importance of the ontology efforts,
but I don't believe that
the efforts are going in the right direction.
LO> Actually, "their" goals (i.e., the Semantic Web folks, Hendler,
> Berners-Lee, et al.) are far more practical/pragmatic than people think.
> In fact, it's exceedingly practical (or practicable, as my old English
> teacher would say).
I agree. But I don't think that what they are doing will achieve
the goals they have in mind.
LO> It's true that there are now model theoretic
> semantics for some of these emerging languages, which I think early
> proponents had to fight for, but now have largely convinced the masses
> on the utility for (one way of agreeing on the semantics of your
> semantic language). But mostly the pragmatists and practitioners and
> implementors are the ones in power on these efforts. We quasi-purists
> suggest as best we can. Pat Hayes, for example, is the prime mover for
> the RDF(S) model theory and is active in OWL (Ontology Web Language), as
> is Peter Patel-Schneider, a language formalist from way back (the
> description logic Classic, etc.)
I agree with Ockham, Peirce, Tarski, Montague, and others that model theory
is an important piece of the puzzle. But the puzzle has a lot more pieces
that haven't been put in the right places.
LO> The main point of the Semantic Web in fact is that there be "these
> suitable tools" John speaks of. Many of us believe we can have and eat
> the cake.
I think so too. But I don't believe that the semantic webbers are using
or proposing to build the right tools.
LO> Where is a standard upper ontology/ies in all of this? Pragmatically, I
> wish we had a better one/s back in the recent days of constructing
> ontologies in the product and service space of business-to-business
> electronic commerce, or in the more remote days of constructing
> ontologies for command and control replanning and decision support, or
> the even more remote days of designing and engineering aircraft. All of
> these efforts I have been involved in, and in each of them, too much of
> our time was spent in developing "upper" ontologies to support our
> "domain" ontologies, which really was our direct task. Why? Because
> everyone of them needed space, time, parthood, connection, etc.,
> concepts and axioms. And too often knowledge-savvy developers who knew
> nothing about the semantic modeling issues involved with such concepts
> had to "wing it" wrt an upper axiomatization in order to focus on their
> domain concepts. Times have NOT changed: this is still the way it is.
I agree with most of what you are saying. And as I said in earlier notes, I
don't believe that Cyc, SUMO, et al. are useless. I would very much like to
use them as *resources* from which I could extract many of the components
and axioms that I would put into my knowledge soup -- and from which I would
build the kinds of theories that are needed.
However, I do believe that the goals that motivated Cyc, SUMO, and similar
efforts are misguided. A single standard ontology would be a disaster, and
the more complete it is, the worse the disaster would be -- if anyone made the
mistake of considering it to be complete, completable, or worth completing.
LO> How many times has a Unit of Measure ontology been invented in this
> world? TOO MANY TIMES. Look at all the object models across the world
> created in UML: they are all reinventing pieces of the upper model and
> doing it much worse than we can do it here in the SUO.
I agree. I would love to have good library of components for space, time,
measures, money, etc. But the emphasis is on *library* of compents in
the *plural* -- not on anybody's particular organization or selection
of components.
LO> So I wouldn't belittle our task. And I definitely wish people here would
> take it seriously.
I take it very seriously, and I've been working on it for years. That is one
reason why it pains me to see how good tasks can go astray.
John