Alex, Bill and John,
That WAY oversteps the reality of “at
least science, technology, engineering, medicine and law”, none of which leave
much basis for asserting that a single meaning is usefully occurring in those
technologies.
Science – among the divergent string theorists, quantum theorists, experimental
theorists, all would find the assertion of consistent meaning to be poorly
based in the current science. Heisenberg would turn over in his grave. Even older
science, e.g. Newtonian physics, is only unambiguous when you have enough
constraints around the problem to filter down to a single interpretation.
But most of science today is empirical, not theoretically rich with
singular explanations. Medicine, data mining, market analysis, psychological
models of behavior, all are rife with disagreement even at the level of whether
the tiny amount of mercury in children’s shots causes disease.
Technology – if there were one unified metric of goodness, why are
there so many different ways of building even common products like computers? Same
heat discharge, power supply life, communications radius, and other metrics are
only roughly predictable using the current (I claim ontological) knowledge of
how SOME things work.
Engineering – this is VERY empirical in nearly all modeling of physical
systems for practical engineering purposes. Few engineering teams can even map
requirements documents into singular designs for any basic set of constraints.
Medicine and Law – try to get consistent opinions from a herd of
physicians and lawyers on anything, even how to treat the common cold (e.g.,
consider Linus Pauling’s vitamin C assertions, the stem cell results analyses,
the health care law), and how and who to sue for damages.
I find the assertion that any USEFUL ontology has to have singular
meanings to be inapplicable to nearly ever discipline with deep studies behind
it. Ontologies are only BELIEVED to be singular in interpretation, but fail
the test in nearly any way that is broadly useful.
So the belief that any useful ontology is singularly meaningful doesn’t
hold water, IMHO.
Going back to WordNet as an ontology, add whatever extensions and
intensions you wish, the antepenultimate English speaker will disagree with the
penultimate one quoting the ultimate one’s statements. So that is perfectly
consistent with the practical realities of ontologies, IMHO.
HTH,
-Rich
Sincerely,
Rich Cooper
EnglishLogicKernel.com
Rich AT EnglishLogicKernel DOT com
9 4 9 \ 5 2 5 - 5 7 1 2
From:
ontolog-forum-bounces@xxxxxxxxxxxxxxxx
[mailto:ontolog-forum-bounces@xxxxxxxxxxxxxxxx] On Behalf Of Alex Shkotin
Sent: Thursday, September 09, 2010
10:36 AM
To: [ontolog-forum]
Subject: Re: [ontolog-forum]
(renamed) Terms with fixed/multiple meanings
Bill: - "Like
everything in AI, it seems, it'll be useful in special cases, but not in
general."
Alex: - Bill, I hope these "special cases" include at least science,
technology, engineering, medicine and law;-)
A lot of thing to do:-)
And not a "multiple meanings" in different contexts is a problem but
possibility and precision of definition.
Alex
2010/9/9 Burkett, William [USA] <burkett_william@xxxxxxx>
John: I understand your
point that a "term" in a formal language (e.g., ontology) should have
a single, unique definition - this allows automated processors to (soundly) do
something with statements in the language.
It is important to point out, however, that this requirement addresses a very
small set of users "out there in web-land" - less than 1% I would
guess. The "semantic web" will never materialize with this
requirement because, simply, a very very large percentage of data-creators
don't have the understanding and won't devote the time/rigor required to create
these semantically precise statements. Most will create their schemas and
ontologies and create their data using their natural language
skills/capabilities/facilities - leading to multiple and evolving meanings.
So, realistically, except for a very small population, "terms"
that are used to name things in web-land *will* have multiple meanings.
We can exclude those undisciplined cases and operate in our own small,
rigorous, well-defined world - but how useful will that really be? (Like
everything in AI, it seems, it'll be useful in special cases, but not in
general.)
As I write this, it brings the question of scope to my mind: in our discussions
here are we ONLY interested in talking about formal ontologies with
precisely-defined semantics that can soundly reasoned over, or are we talking
about the "semantic web" (or "semantic enterprises") in
general where, presumably, we can evolve to a point where processors can do
something will all the data "out there in web-land"?
Bill
-----Original Message-----
From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx
[mailto:ontolog-forum-bounces@xxxxxxxxxxxxxxxx]
On Behalf Of John F. Sowa
Sent: Wednesday, September 08, 2010 4:04 PM
To: ontolog-forum@xxxxxxxxxxxxxxxx
Subject: Re: [ontolog-forum] Semantic Enterprise Architecture -
Interoperability?
David and Doug,
DF>> a Semantic Web needs ontologies of terms with fixed meanings
DE> Is this saying that a term (word, phrase, acronym, abbreviation,
> whatever) can only have a single meaning?
We must always distinguish the names of relations and instances
in any formal language from the words in any natural language
that is being mapped to that formal language.
DF used the word 'term' for the symbols in some formal language.
Those symbols should have unique definitions.
DE was talking about the words used in some NL that is being
mapped to the symbols of some formal language.
The names used in the formalism should never be identified
with the words in the NL -- even when their spelling happens
to be similar.
John