Sunday, December 18, 2005
Lattice of ontologies
Paul
said:
> I will also be looking at ISO 13250
(Topic Maps) and
> other standards where concepts are
expressed without
> logical apparatus.
John
said:
Topic
Maps do incorporate a version of logic, albeit
not
a very expressive version of logic.
English
incorporates a very, very rich version of logic.
In
fact, *every* version of logic that any logician has
ever
invented corresponds to some subset of English or
other
natural language.
******
John,
What
you are saying just does not make sense, to me and to others.
First, the desired use of Topic Maps is because there are
complete use paradigms that do not depend on logic , and thus there is not the
types of limitations to logic that we have recently talked about. Recently you have acknowledged the problem
with first order logic, and over and over again you have pointed out the
incorrectness of the Tim Berners-Lee Layer cake and the RDF standard.
Right?
You know this, and you know that topic maps are currently
used in many instances to manage data via sets of concepts. In a very real way, the OMG architecture is
very close to a topic map, in that a set of virtual elements are external to
all data base data models and this "topic" is used to map any data
element in any of the federated databases.
The main real difference is that Topic Maps is an open standard and OMG
is a business model.
I did a complete paper on this in 2003
http://www.bcngroup.org/admin/technologyReviews/InformationalLatency.htm
and yes I was trying to figure out then, as now, why the
industry is allowed and supported in producing almost no new value while they
take ownership over simple concepts.
You and the others who are active now in this ONTAC discussion
can pretend as if I have not raised quality concerns and have offered paths to
address these concerns. But this is why
Penrose and I and others talk about the "Emperor who had no clothes". You are not recognizing the elements in the
discussion that you can not deal properly with. Why? Is it because you
cannot see, or because you do not want to recognize that you see the facts
about business behavior in government IT procurements?
You talk about a lattice, but will not engage in defining
a smallest element. Why?
Many are pushed away from the Semantic Interoperability
Community of Practice working groups because there are problems with
preservation of errors. It is a
frontal lobe thing.
see
Levine, D. & Prueitt, P.S. (1989.) Modeling Some Effects of Frontal
Lobe Damage - Novelty and Preservation, Neural Networks, 2, 103-116.
You
and Chris and Rick need to get beyond this rejection of anything novel.
As I have suggested, the issue of preservation of
behavioral errors is caused by one thing and one thing only. This is the exercise of control by the
consulting IT organizations so that they can do their thing - which is to make
money. The observed fact is that almost
nothing else is going on, other then the consulting IT organizations making
money, from what is a religious like advocacy of W3C.
Like Baptists will exclude Methodists, the OASIS is
excluded from the ONTAC working group because the OASIS thinking leads into new
approaches to problems that are controlled by some set of political
relationships between W3C groupies (like Rick) and government agencies (like
the Office of Secretary of Defense).
Check on the details.
There is nothing deemed as yet illegal about these relationships. But the outcomes border on waste fraud and
abuse. I have in fact started two
investigations on this issue. And when
there is a bald face behavior like your last message, one has to ask the
investigations to look into systemic causes, root causes. John, are you part of the problem?
****
The class of limitations of logic is not the only
issue. It is the complexity that comes
because the W3C and contractor community has not settled on some minimal
logical core such as what is discussed in the paper that Rick pointed us to.
The problems that I have with the Object Data Model, ODM,
is
·
presented as if there is something complicated and
obscure
·
that it is there are claims and suggestions that are difficult
to access intellectually, and that has lots of little proprietary (or slowly
changing) things like ABoxes and TBoxes.
Since I am opening myself up to criticism, let me make a
specific point.
Figure 1 is labeled
"The primary components of a typical
knowledge representation system based upon description logic."
What happens here is ownership. The authors, and the OMG, want not only for there to be a
recognition that a minimal set of logical operations need to be carried out,
which everyone has to agree with, but also the noose of their camel to be under
the edge of the tent. This is merely dishonest. We do not need a camel, we need clear simply
expressed core logic that is not tied to theorem proving. Why?
Because theorem proving has been proved to be un-resolvable (Church,
Turing, Penrose) and because normal everyday web services (as envisioned by the
needs of government) DOES NOT NEED THEOREM PROVING.
The need, related to having a description logic is simple;
negation
intersection
union
the existential, quantifier, implication, inclusion,
definition for all, number restriction are not always needed in cases where one
is simply retrieving data based on type, properties, attributes, relationships. These are for theorem proving exercises.
These things can be put together and have been put
together. So those that want to spend
their life on theorem proving are welcome to do this, BUT NOT ON GOVERNMENT
MONEY.
The minimal need is to have defined negation, intersection
and union. This is present in all
computer languages.