Saturday, November 26, 2005
Center of Excellence Proposal
à
ONTAC stands for Ontology and Taxonomy Coordinating Working Group
It is a working group of
Semantic Interoperability Community of Practice (SICoP)
Communication from Paul Prueitt made
to the ONTAC forum in response to a question.
Regarding the development of ontology hubs
Examples:
First, we should state that an example is already an abstraction that produces mental/linguistic categories, and that the abstract process was in each case what I am referring to. But examples are plentiful.
A chain of events in biochemistry might involve a macromolecule (complex of proteins) that has the role of a specific catalytic environment. This macromolecule might have just transformed out of materials used in a previous reaction where parts of the macromolecule were lost and parts gained (due to the needs of the environment and due to internal properties of the composite (whole). So there is a local history of the materials and a more global history of the environment (say the reactions of the skin to being cut open).
Suppose we have s(1), s(2), s(3), s(4) as a temporally ordered set of states of a specific physical macromolecule. Suppose that we have knowledge of a particular environmental condition that co-caused the state s(3). Examples of environmental conditions are:
On set of flu
Consequences of impact injury
Each of the states, s(i), i being an index, is a state of a reaction chain which is produced (on a regular basis) by cellular mechanisms. The cellular mechanisms themselves are co-produced by gene expression, etc.
The environment certainly extends out beyond the physical location of the macromolecule. The “non-local” influence is expressed using (actual electromagnetic) fields (such as caused by water in the molecule). As elements of a compositional process come within a certain distance there is a field interaction leading to conformational (protein folding) activities. Degeneracy allows a selection from what is available, in order to satisfy the "function" that the environment needs to fulfill (again the reference to Edelman’s work). That function is determined by several things; including global processes.
An example of global processes is the physical support for individual awareness of certain things in a human’s memory. Layers of event reactions are involved. In each case there are global field interactions as well as local (kinematic) reactions. In this case, the event structure is a chain of mental event (as in thinking). “How thought follows” is far more complex than classical logic would have us believe. The same is true for business and government interactions. The complexity can be addressed, but only by letting go of logical constraints, WHERE these “axiomatics” can be shown to be the co-causes of non-interoperability in IT systems.
One MUST say that the environment co-caused the state s(3) of the macromolecule. How does this happen? An answer might be that there is a templating of environment to local protein reactions by the "needs" of the environment. So in each case, s(i) (where i is a ordering index) is a representation of the state of the macromolecule in a reaction chain which is produced (on a regular basis !) by cellular mechanisms.
For more on neural mechanisms involved in production of mental states see
also see
Levine, D. & Prueitt, P.S. (1989.) Modeling Some Effects of Frontal
Lobe Damage - Novelty and Preservation, Neural Networks, 2, 103-116.
Levine D; Parks, R.; & Prueitt, P. S. (1993.) Methodological and Theoretical Issues in Neural Network Models of Frontal Cognitive Functions. International Journal of Neuroscience 72 209-233.
for some work on a neural phenomenon where activity in "gated di-poles" (ie neural assemble functional structures) leads to changes in how next states will be processed by biological neural networks. In these two papers, some work is reported on a neural phenomenon where activity in "gated di-poles" (ie neural assemble functional structures) leads to changes in how next states will be processed by biological neural networks. The Stephen Grossberg category of artificial neural networks (thousands of articles and hundreds of books/conferences) uses the "gated-dipole" model (a differential equation) extensively to produce a search leading to pattern recognition.
The above would be an example of a metabolic reaction chain that I, and others, would like to model using a set of relationships between ontology entities. Half of my dissertation (1988) was on mapping between discrete switching networks and reaction models using continuous differential equations. This theory is called “homology” and has many outstanding open problems. Perhaps these problems, as currently defined, are unsolvable problems. It is my claim that understanding how to re-state these problems or to by-pass them altogether is one key to future semantic web / anticipatory web performance. I continue to be concerned that paradigms acceptable to computer science builds resistance to approaches that recognize these real problems.
Supply chain analysis has been around for many years, developed in agriculture, and the issues are very much the same. So a set of relationships and specification of concepts that would support templating biological reaction networks would also work for real problems in supply chain modeling.
This templating approach to "process models" is advocated in hundreds of papers, including
"Interaction Protocols as Design Abstractions for Business Processes"
by M Desai et al (special issue on interaction and state-based modeling -- in press I think). I am sure that many in this forum are aware of this work.
I referred to this paper in a previous communication to the forum.
If I have John’s suggestion correct, the detail in each “ontology domain’ would not be shared, but there is something that can be shared.
In my examples, it is possible to not have any logic at all and no "microtheories" or mechanisms for inference. Merely language in the form of controlled vocabulary and knowledge of what the terms might mean in various references to process chains. One needs a set of commonly useful templates and standardization regarding data formats.
The “interaction protocols” are shared and underlying data formats are shared, but NOT at the same time, not in the same way and not by the same standard. The sharing has to be responsive to use patterns that ordinary people understand.
The point that I have been making is that reality is poorly modeled using the types of logics that John talks about (with axiomatics being imposed, as with many OWL ontologies). His suggestion that hub ontology have restrictions on the use of logic, is a suggestion that my work supports.
****
****
You asked this same question "MW: Can you give an example of what you mean?" twice more. The other two were both about stratification.
The nesting problem is one that is well discussed in the biological modeling community. The first step in exposing this literature's point of view is to realize how useful biological taxonomy are; and then how limiting taxonomy is when one gets into metabolic reactions. The metabolic reactions can be "abstractly treated as species" and one can try to construct a taxonomy. But this efforts fails very quickly (for reasons that might be similar to the failure of the ontology community to develop a upper taxonomy for business/government).
Now I have to ask you if I have revealed enough of this so that you remember reading about this type of things (nested class-subclass specifications) in various books (many exists about this).
Stratified theory suggests that organizational levels develop as a natural process of separating real energy-material interactions (so that my cells do not interact with your consciousness). The tri-level architecture that I write about in my work, suggested that one cannot solve the task of constructing a global taxonomy because of a type of relativism... everything is ontologically relative to itself.
Everything has a set of causes and is located within an environment. In the specific environment, to some event, some but not all (possible) raw materials are available to the living process that is involved in the event. The living process is aggregating material (even material like quantum material in reactions like the conformational changes in microtubule (Hameroff and Penrose), under the additional constraint of the affordances (J. J. Gibson’s use of the term) of the environment. .
-----Original Message-----
From:
ontac-forum-bounces@colab.cim3.net
[mailto:ontac-forum-bounces@colab.cim3.net]On
Behalf Of West, Matthew R
SIPC-DFD/321
Sent: Saturday, November 26,
2005 2:47 AM
To: ONTAC-WG General
Discussion
Subject: RE: [ontac-forum]
taxonomy of relationships
Dear Paul,
I don't really follow what
you are saying. See questions below.
Regards
Matthew
> -----Original
Message-----
> From:
ontac-forum-bounces@colab.cim3.net
>
[mailto:ontac-forum-bounces@colab.cim3.net]On Behalf Of Paul S Prueitt
> Sent: 25 November 2005
22:41
> To: ONTAC-WG General
Discussion
> Subject: [ontac-forum]
taxonomy of relationships
>
>
>
>
> I feel that the
discussion between Barry and Matthew is am
> important one.
>
>
> What I would like to
see is some senareo developed where
> status changes in
> the structure (parts
within a whole) and the function (the
> behavior of the
> whole) is clear. Specifically I would like to see how a chain
of
> biochemical reactions
might be modeled.
MW: Can you give an example
of what you mean?
>
> I know that Barry can
identify such an senareo and use his "may I say
> 'stratified'"
taxonomy of relationships. But, there
is a
> problem that John
> has referred to, and
which we are all aware; and that problem
> has to do with
> nesting of classes when
the object of investigation is not an
> engineered
> system.
MW: Can you give an example
of what you mean?
>
> I have seen how some
have addressed this nesting by always
> having three
> levels, the middle one
being the precise one, and the lower
> one being the
> substances that are
aggregated to produce the middle one.
> The upper one is
>
"environmental". Rough set
modeling of economic
> transactions fits into
> this
"tri-level" architecture.
MW: Can you give an example
of what you mean?
>
> Barry, I am not
"thinking" as you and your co-authors do; but
> I do believe
> that you (using the way
you think about these relationships) can model
> biological event
formation and changes in behavior with what
> you have. I
> expect that I could
then "see" how you have done this.
>
> The
"axiomatization" of something, in this case, is highly
> problematic.
>
>
>
>
> The domain that Matthew
is working with really is
> reductionist, as are
all
> "things" in
the world of
>
> large engineering
artefacts that has been developed by
> the Oil and Gas and
Process Plant industries
>
> and thus I feel that
the type of degeneracy (again as Edelman
> this term)
> seem in biochemical
reaction chains will not (by design) be
> seen in large
> engineering artefacts.
>
>
> Matthew, you said:
>
> "The kind of
inference
> you are looking for is
quite core to the requirements of such
> an ontology.'
>
> and I guess you can see
that I am making an ontological
> distinction between
> the domain of biology
and the domain of large engineering artefacts.
>
>
> Your comments, both?
please
>
> With respects...
>
> Paul
>
>
>
>
>
>
>
_________________________________________________________________
> Message Archives:
http://colab.cim3.net/forum/ontac-forum/
> To Post:
mailto:ontac-forum@colab.cim3.net
> Subscribe/Unsubscribe/Config:
>
http://colab.cim3.net/mailman/listinfo/ontac-forum/
> Shared Files:
http://colab.cim3.net/file/work/SICoP/ontac/
> Community Wiki:
>
http://colab.cim3.net/cgi-bin/wiki.pl?SICoP/OntologyTaxonomyCo
ordinatingWG
_________________________________________________________________
Message Archives:
http://colab.cim3.net/forum/ontac-forum/
To Post:
mailto:ontac-forum@colab.cim3.net
Subscribe/Unsubscribe/Config:
http://colab.cim3.net/mailman/listinfo/ontac-forum/
Shared Files:
http://colab.cim3.net/file/work/SICoP/ontac/
Community Wiki:
http://colab.cim3.net/cgi-bin/wiki.pl?SICoP/OntologyTaxonomyCoordinatingWG