Re: SUO: Papers submitted to the IJCAI SUO workshop
Nicola,
I agree with you that we really have to spend some time
evaluating the different ontologies, methodologies, and
fundamental distinctions before trying to merge everything
into one big pot.
I also agree that it is important to produce papers that
summarize our positions in a systematic way and to make them
freely available on the web. I have put the main points of
my position into my KR book, but I have also been adding
more material (some from the book, some based on the book,
and some new developments) in new papers on my web site:
http://www.bestweb.net/~sowa/ontology/
However, I also believe that we really need some time to get
together and hash out details. The last the time we had a
chance to do that was in Heidelberg in 1998. We made a very
good start on the issues, but we didn't reach a conclusion.
Papers alone aren't enough, because it is too easy to talk
past one another in separately published papers.
I read your new PDF file, and it still has many of the same
problems that we have discussed before. Just having you
repeat your statements in your papers while I repeat mine
in my papers doesn't help us resolve the issues. Let me list
some of the issues that need some serious discussion:
1. Status of objects and processes. I believe that Whitehead's
process philosophy, which treats objects as "recurring
event types" is the approach that is most consistent with
both modern science and the requirements for supporting
natural language semantics. I gather that you don't
agree with that position, and I believe that we should
spend some time discussing it.
2. Identity. We had a disagreement about the role of identity
and the use of identity conditions in ontology. This is
another fundamental issue that is closely related to point
#1 about objects and processes.
3. Dependence. You have been using Husserl's dyadic
distinction of dependent and nonindependent entities
in a way which I have claimed is merely the first two
parts of one of Peirce's triads. The third category,
which Husserl did not recognize, is the mediator which
brings the first and second into relation.
4. Agents. In your ontology, you leave out the agent's
intentions, purposes, attitudes, etc. In most cases,
the agent enters into the picture through the third
argument of the triads. If you leave out that argument,
you cannot even talk about relevance, importance, goals,
plans, and anything that comes to grips with meaning.
5. Modality. In your axioms, you have modal operators,
which are not further analyzed. As I have pointed out
many times, including Ch. 5 of my book, those boxes and
diamonds don't explain the reason why something is
necessary or possible. I have recommended that Kripke's
possible worlds semantics be refined by going to Dunn's
semantics, which explicitly states the reasons why
something is necessary or possible. This is closely
related to point #4 above, since in many cases, those
reasons can be related to the intentions of some agent.
To summarize: Yes, we should write papers that present our
positions as clearly as possible. Yes, we should develop
our ontologies in some detail to show what their implications
and applications may be. But we also need to discuss these
issues in detail so that we can benefit from one another's
insights, suggestions, contributions, and criticisms.
John