It makes sense to resuse as
much ontology work as we can and define "applications" of the target ontology
to provide a goal and identify current weaknesses (such as pointed out by
Bruce Smith and John Sowa). After all years ago the field had already
discussed the challenge of
type
hierarchies of one ontology split sthe concept Object into Physical-Object and
Abstract-Object, but another decomposes Object into Decomposable-Object,
Nondecomposable-Object, Conscious-Being, and Non-Conscious-Thing.
A simple way, perhaps too simple
a way, of doing that is to use the 3 level approach to ontology
mentioned, agree on some questions that the ontology will address say at the
mid-level and then work to develop a top level for this.
On a more specific note, Pat mentions that
we could us " an existing one or one that is constructed from
existing ones". I'll post a basic article Two
approaches for ontologies building: From-scratch and From existing data
sources by
Djamal Benslimane, Ahmed
Arara, kokou Yetongnon Faiez Gargouri, Hanene Ben Abdallah
Gary Berg-Cross
EM&I
Potomac, MD
I have sent a request for clarification of her first two comments
to
Denise. My responses to the last 3 are below;
(01)>>
(3) clearly state what the end game of an ontology is (there are
of
course
multiple end games, but we need to begin to at least define
what some
of them
are - otherwise they are motherhood, apple pie and
everything in
between);
(02)There are several
goals for the ONTACWG, but the more technical issues
depend on adopting a
Common Semantic Model (COSMO), which would be an
ontology -- it could be an
existing one or one that is constructed from
existing ones by the
ONTACWG. The purpose of the COSMO would be to
specify the meanings of
domain-specific terms and relations using the
COSMO so that applications
will be able to interpret those terms
consistently for logical inferencing
purposes. This is not exactly
"motherhood", as somewhat less than a
majority of people understand
what a precise logical definition of a
semantic relation would look
like. Funny thing you should mention
"motherhood". My slide 7 uses
that as an example of a simple logical
specification of a semantic
relation.
(03)>> (4)
distinguish between tools and techniques that can be used to
build
an
ontology, and begin to identify where these tools and techniques
are
best used
in the development of a robust ontology model;
(04)There are several ontology-building tools available. The
subWg most
concerned with actually building an ontology will be the COSMO
WG and
we will surely discuss what tools we think will be useful.
Did
you have in mind specific tools that you would recommend?
(05)>> and (5) review what work has already been done that is
not labelled
'ontology'
per se but actually does move towards the end
game.
(06)One main purpose of the "Coordinating" working group is
precisely to
try to attain a global view of all work done on
knowledge
classifications systems, which include, as mentioned in the
charter,
ontologies, taxonomies, thesauri, and graphical representations
such as
UML, CMAP, Topic Maps, and MOF specifications. It also
includes not
only work that is completed, but to the extent possible a
listing of
work that is in progress so that unintended duplication of
effort on
the same topic can be avoided. We want to maintain one or
more
resource pages on the ONTACWG Wiki with pointers and perhaps
also
descriptions of that work. We are requesting all members of
ONTACWG to
post references to any work they are aware of -- finished or
ongoing --
to our "PointerPage" --
(07)http://colab.cim3.net/cgi-bin/wiki.pl?OntologyTaxonomyCoordinatingWG/PointerPage
(08)Or, you can send it to me and I will collate and try to
eliminate
duplications. Such references will, I think, help achieve
the point 5
that you raise. If anyone is willing to actually do a
comparative
review of the available resources, or refer us to such a review
already
done, that would be a great contribution.
(09)Pat