John, All (01)
On 13 Feb 2013, at 15:56, John F Sowa wrote: (02)
>
> Summary: The distinctions an ontology requires are determined by its
> purpose. Making distinctions that are irrelevant to the purpose can
> decrease its generality and interoperability. Therefore, the quality
> of an ontology should be measured by its *relevant* distinctions.
>
> John (03)
Amen. (04)
I missed the presentations, sadly, but will look forward to the audio recording
when available. My own thoughts echo Johns (05)
… and I might add for, that the decisions depend on the consequences for those
applications, which can often be expressed in terms of what sort of questions
the system is supposed to answer. (06)
Ultimately, assuming we are building ontologies in order to build better
information systems, then the ontology must contain the entities about which we
wish to hold information. (07)
Eg If the system is purely historical, whether the future is linear or
branching is irrelevant. If a system must answer questions about hypotheses,
then hypothetical entities are almost certain to be required. If most of the
entities are abstract - e.g. intellectual property - then non-material entities
will be required. If experiments, protocols, plans, etc. are an important part
of the information, then a means of representing plans and their relation to
the actual events that are said to follow or deviate from them will be
required, … … (08)
...additional points in response to the slides but in ignorance of the
conversation... (09)
It seems to me that there is an important missing level in the hierarchies so
far described -- "upper domain ontologies" -- the major categories in this
particular domain - that probably needs to be designed or at least sketched in
conjunction with any top ontology and which is very much a part of the
architecture. If you look at biomedical ontologies outside of OBOFoundry, the
"upper domain ontology" is often fairly well specified, but the upper ontology
proper is usually not. In biomedicine, upper domain ontologies usually
consist of a few tens, possible a few hundred of entities, but the ontology as
a whole may have tens or even hundreds of thousands. (010)
A distinction that I am surprised not to see is between first order and higher
order knowledge. If I say that "Placental mammals are defined as those mammals
that develop placentas as a means of supporting their embryos", that is a
statement about all "Placental mammals". If I say that the distinction between
"Placental mammals" and "Non-placental mammals" was first made by Wallace, that
is a statement about the categories. For some systems, such higher order
knowledge is essential; for others not. (011)
Finally, again a plea to be explicit about where the ontology - the definitions
and other necessary information about the entities in the system - ends, and
broader knowledge representation begins. This is likely to be an important
factorisation line in practice. The technologies that are good for handling
definitions are not necessarily appropriate or even capable of handling other
sorts of information - e.g. contingent statements, strengths of association,
probabilities, if…then… statements, complex pathways, etc. let alone ground
facts about individuals. (012)
Regards (013)
Alan (014)
-----------------------
Alan Rector
Professor of Medical Informatics
School of Computer Science
University of Manchester
Manchester M13 9PL, UK
TEL +44 (0) 161 275 6149/6188
FAX +44 (0) 161 275 6204
www.cs.man.ac.uk/~rector
www.co-ode.org
http://clahrc-gm.nihr.ac.uk/ (015)
_________________________________________________________________
Msg Archives: http://ontolog.cim3.net/forum/ontology-summit/
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontology-summit/
Unsubscribe: mailto:ontology-summit-leave@xxxxxxxxxxxxxxxx
Community Files: http://ontolog.cim3.net/file/work/OntologySummit2013/
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2013
Community Portal: http://ontolog.cim3.net/wiki/ (016)
|