Sunday, January 29, 2006
Generative Methodology Glass Bead Games
possibility of complexity arising in grid computing
[368] ß [comment on four issues
(Richard Ballard)
[367] ß [Four Issues facing
Ontological Modeling (Paul Werbos)
[371] ß [Discussion about founding the
Second School (Paul Prueitt)
[151] ß [Comment by Paul Werbos on
invariance of natural laws
[152] ß [Continuing discussion
between Peter Krieg and Judith Rosen
Communication
to part of the SOA Blueprint Technical Committee (at OASIS) à [144],
Communication from Paul Prueitt
http://xml.coverpages.org/udef.html
is on semantic diffusion
UDEF is something that we have supported and agree with:
http://www.opengroup.org/udef/
http://www.opengroup.org/udefinfo/AboutTheUDEF.pdf
but is not supported, for example, at ONTAC. (Can not even be discussed.)
The problem at ONTAC is that the group will not support a multiple ontologies solution having some type of selectionism producing metrics on each "common" ontology.
Several small groups are suggesting that the metrics should aggregate while the common standard/ontology is fixed to the point where the metric says "all future contractual agreement to use this standard/ontology will no longer be based on "this" version." Old contractual collaborations will be notified and the targeted version is selected to be discounted... and a period of discussion will occur.
The methodology that we are proposing can be justified based on the Rosen definitions of entailments, complexity and process. The methodology can be implemented using existing technologies. The methodology maps nicely that what is understood about natural language formation, use and modification.
This is an open system having reinforced contractually based stabilities. The term "stratified" comes to my mind (for some reason).
The solution to semantic diffusion is confounded by the standard ways of thinking.
Breanna Anderson (SchemaLogic) solved the problem by creating a reconciliation of terminological differences methodology and supporting technology. This has been very successful inside of a large pharmaceutical organization, we know from first hand knowledge. But somehow the market cannot get beyond the demand for precision, exactness, and deterministic universal identifiers. The SchemaLogic solution does not address how reconciliation itself can become open in real time, since the existing hash container has to have all required information at run time.
Our solution does address how reconciliation can become open in real time, by creating two scales of category formation. One scale exists at the substructural level via a study of observed invariance, and one scale exists at the functional/behavioral level reinforced by human-to-human intentional contractual agreements.
“Semantics” is pushed outside of the “engines” (actually finite state machines) where structure is made commonly available to contractually agreeing partners. The “semantic diffusion” problem is managed by humans as we have done for centuries. Within the contracted finite state machine, one has a unique identifier for every “thing”. What one does not have is the illusion that the machine and external reality is “sameAs”.
If the contractually reinforced functional/behavioral level is not producing proper metrics, human can step in and make a new contract based on real time information being developed at the substructural level. Thus a novelty detection feature is possible. (Katrina was understood to likely cause the damage it did, but the novelty of this situation could not be processed fast enough to allow informed response.)
Again, the difference between run time and design time that Sandy Klausner mastered in his development of the mythological “CoreSystem”.
Breanna essentially has a hash table with containers having a representation of all possible points of contextual differences for a word or phrase. This technology/methodology is also in NdCore technology Prueitt helped develop for Applied technical Systems as well is in my Orb technology. Ballard's Mark 3 system may bring this differentiation of all possible meaning into an n-ary processor. But for now, the concept of a limited set of contextual meaning for a term or phrase (or relational element) is not available in the market place.
We may be able to do something about that, using a OASIS standardization process extending and supplementing FERA (Federated Enterprise Reference Model), already an OASIS standard.
So the disambiguation becomes "situational". The information (placed into the container from past work) acts to allow a real time radical reduction to the cost of ambiguity.
The solution that the Soviets worked on (Pospelov and Finn) was to create a measurement process (similar to what Goran talks about) to judge when the computations are producing states that are no longer valid. At that point there is a recreation of the situational logic (using Mill's and QSAR).
This re-creation can be guided by an orchestration process using patterns, as we have outlined in the short PowerPoint. In some cases, the automation fails to produce measured correspondence to "reality" and a human must be consulted.
There is a lot of elegance to solutions to what is called "semantic diffusion" - but in almost all cases the solutions are not found because of the misuse of language itself. Already "semantic diffusion" has diffused the meaning of the word "semantic"... There are two causes of this.
1) the attempt to abbreviate a longer expression such as "the development of an impedance mismatch between the model and the real world"
2) a false understanding of the problem that needs to be solved.
The short paper
http://semanticweb2002.aifb.uni-karlsruhe.de/proceedings/Position/gupta.pdf
outlines the value of a Semantic Web, that others feel cannot develop based on the current W3C paradigm.
This does not mean that value has not be produced so far from the SW concepts, specifically the concepts related to unique identification.
http://informationr.net/ir/7-4/paper136.html
is quite good in defining what unique identification would mean to an open system. This is from a librarian’s point of view.
"The experience of librarianship in controlling large, heterogeneous collections of bibliographic data suggests that the real obstacles facing a semantic web will be logical and textual, not mechanical. Three issues are explored in this essay: development of a standard container of information, desirability of standardizing the information hosted by this standardized container, and auxiliary tools to aid users find information. Value spaces are suggested as a solution, but dismissed as impracticable. The standardization necessary for the success of the Semantic Web may not be achievable in the Web environment."
The solution that my group has in mind, and represented in the PowerPoint, has two scales of interaction. Legal contracts select "value spaces" and parties agree to create a "closed system" for a period of duration and for the purpose of data and process interoperability.
The processes involved in defining
1) invariance
2) patterns
3) blue prints
are open processes that are harvested to create the selection of a closed system (which is reinforced by a intentional contract between parties). Later on, a different partnership will select a different closed system, etc.
Bottom line: Life really is complex in a way that computer programming cannot model. The reduction of reality to a simple system, a model, is possible given current technology. But this reduction should be considered temporary, because the world changes in ways that cannot be precisely predicted.