Tuesday, April 11, 2006
Anticipatory Web discussionà [home]
Anticipatory Web Challenge Problem à [link]
Community Centric Service Methodology Glass Bead Games
Miko
I thank you for the conversations about SOA Blueprints and for your work on the TC.
This note is about a stratified methodology for blueprint generation.
I ask for comment from Drs SOWA, Ballard and Werbos, and from David, Andy and Andrew... all whom have some background in my proposals and in this field generally.
I am making this communication following a long personal practice of explaining a theory about how information technology might work, if based on specific scholarship from systems theory and related (linguistics, evolutionary psychology, neuro-cognitive and quantum cognitive neuroscience). We call the approach "second school" so as to remind the reader that this is not what we have now. What we have now is "first school", and first school has many uses within the second school. However, the second school does not assume or assert that the computer programs can produce an induction of meaning.
The second school asserts that the induction of meaning occurs within the pragmatics of real time interpretation (by humans).
Bottom line: Humans are necessary if the computer/human system is to be "intelligent".
The scholarly grounding is not accessable to most individuals involved in technology start-ups nor to most government program managers. It is however, plainly available to be learned from open scholarly resources.
A Roadmap for Semantic Technology Adoption covers some background
http://www.datawarehouse.com/search/?FREETXT=Prueitt
as does quite a number of BCNGroup / Ontologystream proposals and design documents.
But let us set all of this aside and start simply.
***********************
Background: Why
there is still a SOA design problem.
UML and the OMG (Object Management Group) have a MDA (Model Driven Architecture) which is used to produce Blueprints
http://www.omg.org/gettingstarted/what_is_uml.htm
for SOA collaboration.
At the link above it is said:
<quote>
"Models vs. Methodologies: The process of gathering and analyzing
an application's requirements, and incorporating them into a program design, is
a complex one and the industry currently supports many methodologies that
define formal procedures specifying how to go about it. One characteristic of
UML - in fact, the one that enables the widespread industry support that the
language enjoys - is that it is methodology-independent. "
<end quote>
However, this language is misleading. Specifically, there is the implied assertion that there is nothing left to do, that the problem (all problems) are solved. UML sets squarely inside the first school paradigm and cannot recognize any aspect of the second school arguments.
The tri-level architecture (designed by myself (1994 - 1999) after Soviet era control theory and cognitive quantum neuroscience) introduces a stratification of measurement about the invariance in situations. Various formalisms apply including quasi axiomatic theory... where formal systems are build up from a set of observed facts without (necessarily) imposing first order logic. When the formal system seems to be acting wrong, the entire structure is discarded and a new set of observations made and new formalism developed. (see Chapter 6 of my book) The "formative" process was specified, and can be re-specified as an OASIS standard (when modified to the SOA context).
The stratification theory itself is mapped empirically to quite a number of real physical phenomenon.. including the mechanisms through which linguists suggest produces speech acts, and cognitive science suggest produces cognition and perception. But the simplest to talk about is physical chemistry, where the lower lever is the set of 98 natural atoms and the middle layer is organic and in-organic chemistry. The upper layer is the set of conservation laws of natural reality.
We do not however, claim the tri-level architecture as being able to produce "intelligence".
What we claim is that the invariance over many instances are clustered together by natural forces to produce (natural) categories and that these categories are then re-assembled into "new" real events (at the level under observation). The tri-level has a lower level where the invariance is measured, and a upper level where templates are imposed due to the "expectations" of the environment. These two levels are operating at very different time scales and independently.
Real time phenomenon, including the decision making process entangles the upper and lower levels to produce what we experience as reality. This is second school. The process compartment hypothesis (Prueitt , 1995) simply states that process compartments (localized coherence) (1) forms, (2) has a stable period where replication mechanisms are involved in preserving system identity, and (3) collapses. This hypothesis may seem exotic but it is just what we observe to be true about every aspect of natural reality.
In correspondence to reality
Contract based SOA agreements need (1) a formative process that takes into account the invariances over many instances, (2) governance and measurement and (3) re-view and re-structuring of agreements after a period of time.
The fact is that UML based web service definition has been around for awhile (even before SOA standards) and yet the core issues related to how community interact (from differing viewpoints) is not setting.
The SOA literature shows that SOA is imposed from the top down by management (with the aid of UML) or from the bottom up from isolated IT activities unified through UML.
SOA needs to be "information centric" as Andrew Townley has suggested, and "community centric" as others have suggested.
This is not done well with imposed UML nor with OWL. However, if UML and OWL and data object IM (like SOA-IM) templates (blueprints) are automatically developed (as suggested in the BCNGroup Roadmap) using the tri-level approach; then the humans can select.... this selection between alternatives by non-IT stakeholders makes all the difference in the world. It is a question of understanding and ownership.
Communities need to inform the "meaning" of information structure ... and this inform process can take place with the aid of process defined by www.wsmo.org and www.businesscentricmethodology.com and others.
But for communities to inform the induction of information structure, there is a need for a formative (automated) process and this formative process should follow the natural science exposed by the tri-level architecture.
The tri-level architecture: a solution that measures the evolution of naturally occurring information structure.
The theoretical foundation
1) exists
2) is introduced in the chapters of my book , along with adequate scholarly references
http://www.bcngroup.org/area3/pprueitt/book.htm
There are formal structures (not description logic) expressible as simple n-ary (co-occurrence patterns)
http://www.bcngroup.org/area2/KSF/Notation/notation.htm
The underlying technology is similar to so called "semantic extraction" techniques used in the intelligence community for the past 25 years.
I may have made some mistakes, so please help me see where.
Paul S Prueitt