4 May
2024
4 May
'24
8:03 p.m.
Bobbin,
I changed the title for this topic. Before discussing any issues of modeling anything,
it's important to start with an example. I suggest that we make this topic an
Ontolog project:
You send us a specification of whatever holonic structures you would like to represent.
Instead of using OWL, I suggest that we use Controlled English for two subsets of Common
Logic: (1) A type hierarchy specified by Aristotle's syllogisms, and (2) Full
first-order logic for the constraints.
Then anybody who prefers OWL can map the controlled English to OWL (if they can). If they
can't, that would show why you had difficulty in mapping your problems to OWL. I also
recommend controlled natural languages (CNLs) for any other languages anybody would
prefer. But we can stick with CE for the Ontolog discussion.
For a tutorial on Controlled English (CE) and its mapping to logic, see Patterns of Logic
and Ontology: https://jfsowa.com/talks/patolog1.pdf . Those are slides that I used for
the first day of a 5-day short course that I taught in 2019. Patolog1 should be
sufficient for an intro to CE. If anybody needs more examples, see Patolog2, 3, 4, or 5.
And by the way, these slides are closely related to my book Knowledge Representation,
which was published in 2000, but they were updated with another 20 years of publications
and collaboration with Arun Majumdar and other colleagues in our VivoMind company. I
discuss some VivoMind examples in various slides, especially Patolog4 and 5.
If you still have any of the notes that Pat Hayes sent you, I suggest that you package all
of them in one PDF together with whatever holonic specifications you are working on. Then
send them to us as just one file. It's better to keep it all together instead of
putting them in a collection of a lot of little files.
And by the way, if anybody has trouble with CE, just look at the examples and make your
best guess. That is how we designed the controlled English for VivoMind and for our new
Permion company. It's very forgiving. It never says "Error". It just
comes back with an English echo and the question "Is this what you mean?" If
not, you can revise it until you're happy with the echo.
My personal opinion about the SW stack: It's legacy software that nobody should ever
need to learn or use. All knowledge representations should be specified in CNLs
supplemented with diagrams. The only people who would have to use other notations should
be those who spent years in designing and/or learning them. Today we have the technology
for supporting that policy.
John
----------------------------------------
From: "Bobbin Teegarden" <teegs(a)earthlink.net>
John, a question re your suggested hierarchies below:
I have been frustrated by owl’s lack of composition. Even UML has composition (and
aggregation, a special case of composition). I have been working with holons and holonic
structures (for contexts and such) and need compositions. Pat Hayes once told me that
there were lots of composition/mereology ontologies out there, just pick one.
Frustrating.
Your suggestions solicited. How would I model complex holonic structures on OWL?
Thanks, Bobbin