Generative Methodology thread

Saturday, February 11, 2006

On using RDF to model web services

Reply to à [163]

It is surprising how we can talk completely past each other, not understanding the motivation at all.

At core, the issue that I and others, bring up is that complicated languages such as CLIPs and OWL are both similar and dissimilar to the formal language of say Hilbert mathematics.

The differences include how one defines what an axiom is. In the case of mathematics there is a small number of axioms that were in specific ways complete, minimal and consistent... up to a point. That point is discussed in the Godel theorems. These Godel theorems open up an understanding that Hilbert mathematics is a tool, that when not used well leads to error. The next step is to consider carefully the works of scholars like Roger Penrose and Robert Rosen.

In studying and contemplating these works it can become clear that that part of reality which is modeled extremely well by Hilbert mathematics does not include biological expression, cell expression, gene expression, social expression. This statement, about the universal ness of Hilbert mathematics, already raises defenses. The defenses are of the "emperor that has no clothes type", because of the strong use of polemics. (One expresses indignation over the suggestion that mathematics (or formal logics) are not "universal". A similar polemic is used in the defense of religions, or a social philosophy such as communism.)

All that is asked here is that the concept of a limitation of strong formalization such as Hilbert mathematics but contemplated. Those of us that point this out are not trying to insult anyone, we are merely trying to point out to the (over very over specialized) knowledge engineer that there are results from other disciplines which have a direct bearing on the W3C, AI , knowledge engineering discipline.

Coherence is how the human brain works. The evidence for this is from Karl Pribram and others in work in the field called quantum neurodynamics. This evidence suggested that electromagnetic field coherence mediated by the connectivity of neurons, comes to reside for a period of time as we think. Compare this explanation about the nature of thought with the notions of logic, seen in the wide perspective of classical thought.

So the analysis that is made is that "logic" has within its own discipline an understanding of the problems of limitations; non-computability and non-formalization and un-decidability etc. And yet inspite of evidence of non-removable limitations to the concept of formalization, some strongly ignore the notion that a formalization of "logic" can be found that plays the same role as mathematics does to engineering, but in the general case.

In the standards committees at W3C and at OASIS people talk about web service orchestration and web service discovery; as if this is a kind of thing that can be engineered and still have fidelity to natural processes involved in human business interactions.

OK, so there is an alternative viewpoint. This viewpoint is that "the knowledge engineers" have not had time or resources to fully explore the notion of a semantic web where web service orchestration and web service discovery is fully enabled, in the general case.

The fact is that semantic web research will continue during your life time and mine. But there are some hard facts to address concerning several issues, each one of which would be strongly rejected as being an issue that should be discussed (here or anywhere).

The behavior is to re-define what I have said; attack that re-definition; and then point out that I stated the flame war. I am not interested in the flame war notion of behavior; only a open discussion between scholars (from different fields).

The strong need for a stable and understandable OWL editor is being weighed against a research community’s interest in continuing a line of research (which some regard as being like cold fusion). Perhaps doing both is what NSF should consider funding. There are clearly benefits to the esoteric research into yet another way to avoid acknowledging limitations that some of us regard as un-removable. This work is fun, to those who have made a profession out of it.

But there are several layers of practical considerations. One being that "the world" is flocking to the illusion that general purpose ontology for everyday types of things "like human decision making over which song to play on the radio" is easily available using Protege:

I quote a message yesterday:

"We are about to get on board with the Semantic Web and creating ontologies within our organization. We want to start out with something relatively small and then work our way up to bigger things.

My question is this. What would be the requirements for setting up a server or work station. Are there minimum hardware requirements? What software would I have to have loaded to get this up and running. We would like to Demo something to show proof of concept. Are there any Database software requirements that would be needed. We want to be able to search and store the results from our queries and then forward our results to someone in the field who needs this information."

The discussion I am causing over this issue of usability of Protege is specifically couched in the question of whether a small OASIS supported information model (having 38 concepts) can be expressed in RDF, RDFS, OWL Lite, OWL DL, OWL Full; or Protege frames. An alternative is the concept ontology using n-aries but without "inferential logics".

This SOA-IM is produced from the XLM world, where it is assumed that a conversion to OWL ontology is merely a matter of someone doing the few hours work. So here end is the challenge.

Can anyone of those who are regular contributors to these two forums make this conversion?