Re: SUO: Multi-Source Ontology (MSO) Draft Ballot Question
Adam,
The distinction you're making is very important:
AP> Another point is that WordNet and SUMO are two very
> different products that one shouldn't mix together.
> WordNet synsets are language constructs, not concepts,
> and the links among synsets are correct linguistically,
> but not philosophically. SUMO on the other hand is not
> language dependent, and defines concepts which are valid
> philosophically, but may not have direct correspondence
> to any particular lexical item.
But as you know, many people, including the Cyc and SUMO
developers themselves, have done a lot of work to relate
their concepts to the WordNet categories. That alignment
is essential for the simple reason that most of the words
that we are using in our speech, publications, and emails
are found in WordNet, but not necessarily in the more
formally defined concepts of Cyc and WordNet. Somehow,
we have to relate our English words to our formal
concepts, and something like WordNet can help us do so.
AP> I fear that putting them in the same representation
> language could cause considerable confusion.
That fear is well founded. But exactly the same confusion
is possible when people see the WordNet categories aligned
to SUMO or Cyc. It is important to note two points:
1. The alignment is necessary for a great many reasons.
2. But we have to draw a clear distinction between the
different kinds of categories.
Where we disagree is over the following point:
AP> WordNet synsets are language constructs, not concepts,
> and the links among synsets are correct linguistically,
> but not philosophically.
The major disagreement is over the words "linguistically"
and "philosophically", which are as vague as the Bible:
you can always find a quotation from the Bible, some
philosopher's writings, or some linguist's writings
to support any claim you want to make.
AP> As an example, take the inference path that is possible
> from the SUMO term Motion to the WordNet synset "motion,
> movement, move" then up the hypernym links to "change",
> "action" and then "act, human action, human activity".
> Through that faulty inference chain, one could conclude
> that any Motion is an intentional human action, which
> of course is false. Hypernym links are linguistically
> valid, but not logically valid.
I agree with everything but the final sentence: the
chain of hypernyms you cite are definitely not valid
linguistically. Chains like that have been criticized
by many linguists, and they cause exactly the same kinds
of errors in linguistic analyses that they cause in SUMO.
I also agree with the following point:
AP> WordNet wasn't intended for logical inference, so if
> you want a product for doing inference, you're better
> off with a product created for that purpose.
Unfortunately, there is no product for doing inference
that has a coverage as broad as WordNet. Cyc is coming
close, but it still isn't close enough. That is why we
need a project that will start with the resources that
are currently available, such as OpenCyc, SUMO, WordNet,
LIS, and others, and develop a new product that corrects
the faults and extends the good features of all of them.
That is exactly the goal that Philippe is working toward
with MSO and the supporting tool set of WebKB. The motion
we are voting on is whether the SUO Working Group should
accept Philippe's offer to give us MSO as a basis for
achieving that goal.
So far, nobody else has provided anything else that comes
as close to meeting the SUO goals as MSO. When I say that,
I am not belittling the work of the other groups, such as
SUMO or OpenCyc, because Philippe is accepting them and
incorporating them into MSO. Furthermore, he is providing
tools in WebKB for documenting every criticism or correction
to MSO or any of the ontologies included in it.
John