John, (01)
Sorry for the delayed response. (02)
The primary, or at the initial, goals of the OOR are to
provide repository for 'good' ontologies, where an 'ontology'
is something represented in one of the standardized
representation languages and has been passed through
the OOR gate keeping (aka quality control) process. (03)
Now this gate keeping process will at least provide some
simple correctness checks. Additional 'goodness' checks
will represent best practices (e.g., inclusion of annotations). (04)
In addition to gate keeping Ken Baclawski introduced the
notion of work flow. The ability to enforce common vocabularies
across a domain are beyond the scope of current requirements
and expectations. However, I could this as specified via
policies and enforced as a work flow. Of course the enforcement
of these more global policies would be 'expensive' (from a
computational point). (05)
One of the architecture principles for the OOR is
modularity. So if we create the specification correctly
and conform to the current requirements and principles,
the OOR should be able to accommodate the addition of a
capability that provides enforcement of a common domain
vocabulary. (06)
Leo Obrst has brought up the issue of a configuration for
ontologies. By this, to my understanding, he has in mind
aspects of ontological architecture, dependencies and
required imports and such. So the notion that an ontology
has to conform to a particular vocabulary could be part of
such a configuration. (07)
Todd (08)
From:
"John F. Sowa" <sowa@xxxxxxxxxxx>
To:
oor-forum@xxxxxxxxxxxxxxxx
Date:
03/26/2011 11:04 PM
Subject:
Re: [oor-forum] OOR architecture & API Workshop-III - Fri 2011.03.25
Sent by:
oor-forum-bounces@xxxxxxxxxxxxxxxx (09)
Tim and Todd, (010)
When I asked those questions, I actually thought that the answers
were fairly straightforward, at least for the first four.
But I believe that these issues (or some variation or extension
of them) need to be resolved at an early stage in the design. (011)
>> 1. What is an ontology?
> --->>> The OOR is not attempting to explicitly define what an
> ontology is or isn't. (012)
Following is my definition, which covers every formal ontology
I've seen: (013)
A formal ontology is a theory, stated in some version of logic,
that specifies the semantics of a set of types and relations. (014)
>> > 2. What is a terminology?
> --->>> Again, I don't think OOR will try to explicitly (or implicitly)
> define this. (015)
My definition: (016)
A terminology is a set of terms (words or phrases) in some natural
language with informal definitions stated in a natural language.
Some relations among the terms may also be specified, such as
'more general than', 'more specialized than', or 'disjoint with'. (017)
>> > 3. How are ontologies and terminologies related?
> --->>> OOR won't explicitly address this. (018)
A formal ontology may be specified as a formalization of
a terminology by relating each term to types and relations
specified by the formal theory of that ontology. Each of
the relations among the terms must also be formalized by
relations defined in the same version of logic. (019)
>> > 4. What is the underlying semantics of an ontology, and
>> > how is it related to the semantics of a terminology?
> --->>> OOR won't explicitly address this. (020)
The formal semantics of any ontology is defined by the model
theoretic semantics of the logic in which it is specified.
The formal definitions should be more precise than the NL
definitions in the terminology, but they should be consistent
with the expected interpretation of the NL definitions by the
people who use those terms. (021)
>> > 5. When the same terminology is related to different ontologies,
>> > the same words may have different definitions in each of
>> > the ontologies. How are the discrepancies noted?
> --->>> OOR won't explicitly address this. (022)
The answer to this question is somewhat more complex. My recommendation
is to allow multiple formal ontologies at different levels of detail
to specify some or all of the terms in the terminology. (023)
This procedure is commonly done in practice because different
people in the same organization may use the same terms in different
contexts with different amounts of detail and different sets of
relations among the terms. (024)
Examples: (025)
1. In a hospital, doctors, nurses, technicians, pharmacists,
patients, orderlies, and administrative personnel may use
the same terms, but with different levels of detail. (026)
2. Different departments in the same company may have some
common terminology, but with varying levels of detail:
Engineering, manufacturing, sales, finance, human resources,
shipping, maintenance, etc. (027)
I discuss point #5 in terms of a hierarchy of theories. For a brief
summary of the issues, see slides 61 to 80 of the following: (028)
http://www.jfsowa.com/talks/iss.pdf (029)
John (030)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/oor-forum/
Subscribe: mailto:oor-forum-join@xxxxxxxxxxxxxxxx
Config/Unsubscribe: http://ontolog.cim3.net/mailman/listinfo/oor-forum/
Shared Files: http://ontolog.cim3.net/file/work/OOR/
Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OpenOntologyRepository (031)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/oor-forum/
Subscribe: mailto:oor-forum-join@xxxxxxxxxxxxxxxx
Config/Unsubscribe: http://ontolog.cim3.net/mailman/listinfo/oor-forum/
Shared Files: http://ontolog.cim3.net/file/work/OOR/
Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OpenOntologyRepository (032)
|