Bob,
Your points seem to underscore effectively
the diverse uses in academic research and business applications of Ontology methodologies and begs for further discussion of the
challenges of quality metrics in various domains.
Some large firms, for example, are beginning
to focus on Policy driven ontologies ( PDO ) as an approach to evaluating
application interoperability as part of an architectural analysis. One of the
obvious drivers of PDO is external regulatory demands (SOX, Basel II, HIPAA,
etc.) which puts continuous pressures on Boards of Directors for adapting to
new contexts.
Thanks for your insight. How does GSX
cope?
Rob Smith
Tall Tree Labs
From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx
[mailto:ontolog-forum-bounces@xxxxxxxxxxxxxxxx] On Behalf Of Robert.Miller@xxxxxxx
Sent: Tuesday, May 03, 2005 8:09 AM
To: ontolog-forum@xxxxxxxxxxxxxxxx
Subject: RE: [ontolog-forum] Re: Semantics
Ron Schuldt writes:
Isn't it
also safe to say that any ontology must select a definition for a given word
and then use it consistently throughout the ontology - regardless the other
possible uses?
Chris Menzel writes:
If I want
to keep using "widget" with my meaning, I can rename occurrences of
"widget" in O1 systematically (with "O1-widget", say) and
the ambiguity disappears.
Chris also says:
But one of the points of building an ontology
is to *fix* meaning and thereby to *avoid* confusion.
I suggest that the point of building an ontology is to
document meaning and thereby enhance understanding. That the
meaning of words/terms/phrases may differ with context
is a given. An ontology should routinely include context
in its architecture, such that it is capable for example of documenting
multiple meanings of a 'widget' along with their associated contexts. Note that I do not suggest that ambiguity
will 'disappear'. Nor do I see that as a practical goal. Since an
ontology that tracks a real world will grow/morph over time, meanings cannot be
'fixed'. On the other hand, inference machines work pretty well when
presented with fuzzy data.
Cheers,
Bob
Miller