Re: looking at the merged ontology ("")
Thu, 27 Nov 1997 15:38:52 -0600

>1. In the ontolingua source, "Individual" is a subclass of "Thing", while
>in the generated indented list the two classes appear at the same level.
>This seems to be a problem of the indented list generator.

Yes, Individual is a subclass of Thing. Note that subclasses
are in general not disjoint; Intangible, another subclass
of Thing, is not disjoint with Individual. Disjointness, where
it occurs, is asserted explicitly in Cyc.

>2. Is the merged ontology just the latest version of CYC upper level plus
>the SENSUS links? In other words, when writing the paper, should we
>consider separately the merged ontology on the Stanford site and the CYC
>upper level as available on the Austin site?

I believe there is no difference between them. We added about
30 terms to Cyc, and hundreds of links to Sensus (of 3 types);
i.e., all that is now part of the Cyc upper ontology. This
is what you get (including those new terms and links) if you
go to our website and click on the ASCII or
POSTSCRIPT download of the Cyc top-level ontology (roughyl
3,000 terms and 3-15 simple axioms/assertions involving
each one) available right there on our home page. The only
difference is that the ontolingua folks renamed a few of our
predicates (e.g., "isa" to "instance-of", which is what
we used to call it; we renamed it a few years ago because we
type this a lot and "isa" is much shorter than "instance-of".
The same for "genls" versus "subclass-of".) But I don't
believe that they removed or added any terms or assertions.

>3. Why so many annoying redundant generalization links, like
>Wood subclass-of Solid-Tangible-Thing
>Wood subclass-of Structural-Support-Stuff
>Structural-Support-Stuff subclass-of Solid-Tangible-Thing
>isn't subclass-of transitive?

Yes, subclass-of [i.e., genls] is transitive. Note
the assertion: (isa genls TransitiveBinaryPredicate)
which is one of the upper-level ontology assertions. So
the short answer is: yes, there are some redundant links;
if you want, remove them. If you want, we'll remove them.

The longer explanation, in case you are curious is:
In order for us to produce this one-microtheory file, from Cyc,
we flattened hundreds of high-level microtheories. In some
of them, some intermediate terms (e.g., Structural-Support-Stuff)
didn't even exist, hence there were direct genls links such
as from Wood to Solid-Tangible-Thing. In other words,
what you are looking at is a sort of 2-dimensional projection
of a higher-dimensional construct which is the Cyc ontology
that we (Cycorp and its customers and collaborators) use.


Dr. Douglas B. Lenat phone: 512-342-4001
Cycorp, Suite 100 fax: 512-342-4040
3721 Executive Center Drive email:
Austin, TX 78731 web: <>