topicmaps-comment message
[Date Prev]
| [Thread Prev]
| [Thread Next]
| [Date Next]
--
[Date Index]
| [Thread Index]
| [Elist Home]
Subject: [topicmaps-comment] Ontologies -- the identification of what "is"
- From: psp <beadmaster@ontologystream.com>
- To: eventChemistry@yahoogroups.com
- Date: Sun, 30 Dec 2001 09:24:40 -0500
I copy
Dick note below for reference in the post to the KMCI. As requested by
David. I also copy to the OASIS Topic Map Comments forum. I do not
want to burden this forum with the direction we might take in eventChemistry and
therefore I use this cross post to invite a discussion at eventChemistry (which
is closed - so you have to ask to be a member...
***
Dick
(Ballard) said:
"one
needs some hint or familiarity with the words or symbols being used -- in
order to confirm that you are "accepting and committed" to the existence
and employment of those ideas"
Paul
->
Yes
this is the problem I have in the development of Incident Management and
Intrusion Detection (IMID) software ontology. The feedback from experts
must be "situated" in order to nominate the proper conjecture relationship, the
scope (period) of the data source, and the work produce of some event graph and
description of the nature of a cluster.
The
"action-perception" cycle also requires, in the best of all worlds, that
the user take action in context of the terms being used by the recognition
system.
***
Dick
said:
"You
might use Roget's to take all the words used in tiling your subject idea or
concept, looking through all the numbered categories in her thesaurus to
pick the one most appropriate. (Roget's always has only 800-1000 categories). As
you accumulate all the ideas you need to define or express clearly, you might
find several that were placed in the same numbered category, but still
are somehow different. Conventionally, you might create then some "kind-of"
taxonomy to further differentiate these "fine grained" subject or action or ...
whatever ...distinctions. Kipfer's list follows relatively well the traditional
content groupings used by philosophers."
Paul
-> Do you have a machine version of these 800-1000 categories? This
would be useful for my work on fable indexing and IP indexing. The notion
of tiling the concept space is the same as the notion of a descriptive
enumeration as used in my SenseMaking presentation while I was at Wizdom
Systems:
ALSO -
AND VERY IMPORTANT -- In the IMID software task, we need to
evolve a system of signs (in the Peircean tradition) that corresponds to
the
1) the
invariance in the data (as found by data mining and treated roughly as
nouns)
2) the
intentions (or inferred intentions) of hackers or defenders (white hats /
black hats) (treated roughly as verbs)
Is
this what you have in mind Dick?
I am
hearing what you are saying about " "forms", independent of subject, e.g.
the Peircean trichotomy [independent (entity), relative (relationship),
mediating (structures)]; Physical/Metaphysical dichotomy; and the
Dichotomy/Trichotomy of Time Dependence [occurrent, continuant, universal].
Kipfer might take these same ideas and fit them into categories of state,
construction, quality, measure, etc."....
but
perhaps (JUST PERHAPS) almost any choice of a good framework will work much
better than any framework in existence (at least for IMID work). We do not
have to wait for some optimal choice, since new the work is perhaps best
characterized as random and blind.
Dick,
and other participants here:
The
opportunity to build a back end ontology for IMID eventChemistry offers some
unique challenges:
1) the
parts of "speech" for invariance detection have to at first be assumed
to not have any subcategories... each atom is an atom with no knowledge of
the role the atoms plays in anything. We have a bag of atoms - each one
having difference relational affordance (valance), but we have no theory of type
as yet.
2) the
human domain experts must use the "make comment" feature in the SLIP Technology
Browser and THEN knowledge engineers such as Drs Murray, Jones, Ballard and I
must review these comments in order to develop an abstract theory of type where
we begin to develop "parts of expression" sub-categories (based on our
experience and on linguistic theory and chemical theory)
The
hypothesis that I have put forward to the CERT (Computer Emergancy Response
Team) community is that a "blind development of theories of natural kind and a
theory of cause" from the way that CERT specialists organize the data and make
comments about the data, will lead to global "Informational Transparancy with
Selective Attention" regarding
1) all
IDS data sources (now)
2)
(and then later) all bit map transactions occurring through vital Information
gateways, networks and single processors.
The
SLIP and eventChemistry Browsers where developed for this
purpose:
Software from Software Systems
International
is one
way to place a sensor system "below" the IDS (Intrusion Detection Systems)
systems so that new exposure can be taken care of faster than the two to three
weeks it now takes to add rules to IDS systems. S. Forrester's work (NMSU)
on theoretical immunology for computer systems is another.
The
"recognition and cataloging of "mediating conceptual structures" -- the
rational frameworks with which we answer questions of "how", "why",
and "what if" and find acceptable certain patterns of causal association and
reasoning from theory" is indeed the very reason why I have found your
(Ballard's) work most refreshing.
The
theory of "mediating conceptual structures" for IMID is likely to be much
simpler than for natural language. This is possibility that OntologyStream
will develop a break through in IMID ontology encoders is high. We are
asking for help from Dr. Jones' group and from some members of the topic map
community.
Imagine a "private language" that evolves so that IMID
agents can talk to IMID agents about the invariance in the bit stream and the
likely intentions of whose humans and human communities that are causing these
measured invariance. This is the concept OSI is working
on.
My
time scale is quite different from Dick Ballard's, in that I have to completely
solve the ontology encoding problem for IMID systems within the month - at least
to the point where the various scholars who interact with me can see where this
is going and to the point where either my former client starts to fund again the
development work or I obtain a sole source contract from an agency or the White
House directly. (And yes, the frustration level about not being able to do
our work due to funding issues is extreme - why shouldn't this bother
us?)
My
recent (actually long term) interactions with the investment and lawyer
community convinces me that reasonable expectation of profit is not sufficient
for most of these folks. They want control... and do not understand what
blind control of innovation feels like to the innovators. (It is not able
profit of success it is about personal inertia and ego).
If
only they would come off of their high horse and work just a bit to understand
what the true issues are. Enlightened management is rare. For the
BCNGroup the long term challenge is clear, in that this enslavement of
innovators by capital has to be removed by structurally dominating the
capitalist (using money) to the role of making a positive return on investment
(something that has not been done very well in the .com era) and away from
intellectual control of something that they care not about and do not have the
specific educational background to address.
Help
is asked for....
let us
make this work?
Don:
An
ontology is simply an enumeration of what "is". On its face, it does not
require some particular level of description or detailing, though as a matter
of course, one needs some hint or familiarity with the words or symbols being
used -- in order to confirm that you are "accepting and committed" to the
existence and employment of those ideas. You accept the
speaker "ontological commitment" or throw certain notions out as error,
fraud, or worse.
In
my ontology paper (enclosed, if you or others have not seen it already), I
talk of two popular ontological references. (1) Barbara Ann Kipfer's Concept
Index in "Roget's 21st Century Thesaurus" (Delacorte Press). (2) The
Pierce-Sowa-Ballard system detailed in my paper, the bulk of which is derived
from John F. Sowa's "Knowledge Representation - Logical, Philosophical, and
Computational Foundations" (Brooks/Cole).
In
that paper, I point out the distinctions in purpose between these two
ontological views. Kipfer's is tied closely to "subject (word) ontologies".
You might use hers to take all the words used in titling you subject idea or
concept, looking through all the numbered categories in her thesaurus to
pick the one most appropriate. (Roget's always has only 800-1000 categories).
As you accumulate all the ideas you need to define or express clearly, you
might find several that were placed in the same numbered category,
but still are somehow different. Conventionally, you might create then some
"kind-of" taxonomy to further differentiate these "fine grained" subject or
action or ... whatever ...distinctions. Kipfer's list follows relatively well
the traditional content groupings used by philosophers.
Now
what Pierce, Sowa, and I are distinguishing is between "forms",
independent of subject, e.g. the Piercian trichotomy [independent
(entity), relative (relationship), mediating (structures)];
Physical/Metaphysical dichotomy; and the Dichotomy/Trichotomy of Time
Dependence [occurrent, continuant, universal]. Kipfer might take these same
ideas and fit them into categories of state, construction, quality, measure,
etc. Finding agreement on an enumeration of concepts is relatively easy,
agreement on a common framework of categorization or abstracting the lot,
is in practice virtually impossible. In speaking of universals, Plato listed a
great many different and relatively independent points of view, take esthetics
(beauty) or ethics (morality), or politics, ....
Sowa
is interested in the categories we detail, in part, for their
relationship to mathematical and natural language notions like logical order
and syntactic form. My interest is tied to conceptual visualization and
universal browser design. From my perspective, the frontier of knowledge
acquisition and tool design today is in the recognition and cataloging of
"mediating conceptual structures" -- the rational frameworks with
which we answer questions of "how", "why", and "what if" and find acceptable
certain patterns of causal association and reasoning from theory. These last
issues are about systems of knowledge representation -- separate
them in your mind from the notion of ontology which has only to do with the
titling and coded uniqueness of ideas, the matching ideas (particularly
metaphysical ideas) between speaker and listener, and a mutual commitment that
these ideas represent a meaningful and near complete way to describe what
"is".
Without meaning to criticize Paul, he has focused on my ontology paper
and work. My work on browsers and semantic networks is still a mystery to
you all. It is there in knowledge representation that we all need to go next.
For most, the earliest opportunity to see the older Mark 2 knowledge
bases may be the Knowledge Technology Conference in Seattle (March
2002). I will be doing Mark 3 browser designs in late spring -- but for
investors and lawyers may need non-disclosure statements before I can say much
about their functionality. For the past decade, Mark 2 is (was) a
publicly promoted and demonstrated system so I will focus there and you will
see enough to whet your appetites
Dick
Yahoo! Groups
Sponsor |
ADVERTISEMENT
| |
|
To
unsubscribe from this group, send an email
to:
eventChemistry-unsubscribe@yahoogroups.com
Your
use of Yahoo! Groups is subject to the Yahoo! Terms of Service.
[Date Prev]
| [Thread Prev]
| [Thread Next]
| [Date Next]
--
[Date Index]
| [Thread Index]
| [Elist Home]
Powered by eList eXpress LLC