[Top] [All Lists]

[ontac-forum] Wordnet: was Semantic Interoperability: Sowa's Collection

To: "ONTAC-WG General Discussion" <ontac-forum@xxxxxxxxxxxxxx>
Cc: deddy@xxxxxxxxxxxxx
From: "Cassidy, Patrick J." <pcassidy@xxxxxxxxx>
Date: Sat, 26 Nov 2005 23:45:38 -0500
Message-id: <6ACD6742E291AF459206FFF2897764BE697BC4@xxxxxxxxxxxxxxxxx>
Just one thought about WordNet:    (01)

As John mentions, some of the upper ontologies have already been mapped
to WordNet (though I have little familiarity with the mappings and have
never evaluated them).  This suggests that WordNet could, to some
possibly minor but useful extent, serve as an existing "interlingua"
which can help point out likely correspondences between classes in
those upper ontologies that have been mapped to it.  I would expect
such correspondences to need cautious interpretations and careful
examination, but it would provide additional data to help us decide to
what extent the existing upper ontologies could be merged, or what
parts would have to be sequestered from each other in microtheories or
contexts as logically incompatible.    (02)

But, given the immense focus recently on medical knowledge bases and
the Federal Enterprise Architecture, I think it best that we focus what
efforts we can muster in the immediate future on the  formalization of
the UMLS Semantic Network, and the FEA-RMO, so that the work we are
doing will be more likely to be relevant and useful to the work being
done by others with a short-term perspective on applying the work they
are doing.    (03)

To that end I have created a Wiki page
(http://colab.cim3.net/cgi-bin/wiki.pl?CosmoWG), where those who are
interested can put their suggestions for alignment of the UMLS semantic
network type hierarchy.  Discussions will also be conducted within the
COSMO-WG, which is open to anyone wishing to participate in the
construction of the COSMO, or to just keep an eye on what is being
done.    (04)

The suggestion that John Sowa made, to build an Unified Framework (UF)
that will contain all of the classes and relations that are not
significantly controversial, is also, I think, a very good idea.  Since
the focus would in that case be on the class hierarchy, it may not take
as much effort as more detailed formalization work, and could be a
practical and very helpful thing to do.  If no one else beats me to it,
I plan to make some concrete suggestions for an actual process to do
this -- but that may not happen for a week or two.    (05)

Anyone think it is **not** worthwhile?    (06)

Pat    (07)

Patrick Cassidy
MITRE Corporation
260 Industrial Way
Eatontown, NJ 07724
Mail Stop: MNJE
Phone: 732-578-6340
Cell: 908-565-4053
Fax: 732-578-6012
Email: pcassidy at mitre.org    (08)

-----Original Message-----
From: ontac-forum-bounces@xxxxxxxxxxxxxx
[mailto:ontac-forum-bounces@xxxxxxxxxxxxxx] On Behalf Of John F. Sowa
Sent: Saturday, November 26, 2005 11:21 PM
To: ONTAC-WG General Discussion
Cc: deddy@xxxxxxxxxxxxx
Subject: Re: [ontac-forum] Semantic Interoperability: Sowa's Collection
ofModules    (09)

David,    (010)

Controlled vocabularies have proved to be extremely
valuable for many purposes, and they come very close
to being specialized ontologies for a particular domain.
So I would certainly support an approach along the
following lines:    (011)

 > Would it be possible to consider something WordNet-like,
 > but rather than containing some 120,000+ words (& steadily
 > expanding at a rapid pace), have JUST the language &
 > concepts relevant to a particular application.
 > I've had some exposure to a sizable company with a
 > controlled vocabulary list of a mere 1500 "terms"...
 > two full orders of magnitude smaller than WordNet.    (012)

However, there is also a need to build bridges between
different domains.  For example, there has been a great
deal of work on standardized nomenclatures for medicine.    (013)

But medicine is such a large, complex field that it becomes
very difficult to limit the domain.  An important part of
the vocabulary is devoted to the anatomy of the human body,
but as you get to smaller and smaller parts, you get into
microbiology, which leads to issues of metabolism and all
the biochemical reactions, which lead to drugs and how
they relate to the biochemistry, which leads to the genes
and how they produce proteins, which are related to the
molecular structure of the drugs, which are related to the
structure of the bacteria and viruses that attack the human
cells, which lead to the biology of all the pathogens, which
lead to all the issues of biochemistry, pharmacology, etc.    (014)

And when you talk about medicine, you also have to talk about
the medical equipment, which involve chemistry, physics,
computers, electrical and electronic engineering, and all the
possible reactions with the human physiology.  And the medical
equipment interacts with all other kinds of equipment, including
the air conditioning, lighting, etc.    (015)

But a major part of any medical system involves the personnel of
doctors, nurses, administrators, and all the paperwork of billing,
supplies, government regulations, HMOs, insurance, etc.    (016)

Where do you stop?  Very quickly your controlled vocabulary
runs out of control.  And that's just for medicine.  You could
do the same thing for the petroleum industry or the auto industry.
And when you get to the government, you have connections to every
business, educational institution, and home in the country.    (017)

In short, I'm all in favor of developing controlled vocabularies.
But they very quickly run into other controlled vocabularies,
which lead to others, and others.    (018)

A dictionary, such as WordNet, or even the OED, is where all
those controlled vocabularies meet.  That's why I said we should
start with something like WordNet, which has been very usefully
combined with many ontologies, such as Cyc, SUMO, and others.    (019)

John Sowa    (020)

Message Archives: http://colab.cim3.net/forum/ontac-forum/
To Post: mailto:ontac-forum@xxxxxxxxxxxxxx
Shared Files: http://colab.cim3.net/file/work/SICoP/ontac/
Community Wiki:
gWG    (021)

Message Archives: http://colab.cim3.net/forum/ontac-forum/
To Post: mailto:ontac-forum@xxxxxxxxxxxxxx
Shared Files: http://colab.cim3.net/file/work/SICoP/ontac/
Community Wiki: 
http://colab.cim3.net/cgi-bin/wiki.pl?SICoP/OntologyTaxonomyCoordinatingWG    (022)
<Prev in Thread] Current Thread [Next in Thread>
  • [ontac-forum] Wordnet: was Semantic Interoperability: Sowa's Collection, Cassidy, Patrick J. <=