ORB
Visualization
3/17/2004
6:37 AM
J reply from the Professor -> .
Professor of Computer Science
Eric Miller
W3C
Ralph Hodgson and Robert Coyne reply from TopQuadrant -> .
TopQuadrant Inc
Over the past five years the claim has been the persistent that RDF (Resource Description Framework) type ontology constructions could and would provide both interoperability and universal application to social, military and business needs.
Of course the claim goes back several decades, or centuries, or at least a version of this claim. This claim makes the assumption that formal deductive inferences are universally applicable in “reasoning” about the natures of the natural world.
I personally have found the TopQuadrant methodology for making a business case (business capability design and realization) to be based on superb social science and on an understanding of complex process. In some of the presentations by TopQuadrant, I saw what I regarded to be a proper and balanced treatment of Topic Maps. Topic Maps, like RDF, supports some degree of formalization of structure. But implicit in the Topic Map standard 1.0 is the notion that reification and inference often occur outside the computer.
Over the past two years, I have been disappointed in the evolution of a TopQuadrant business case supporting universality of RDF, OIL (Ontology Inference Language) and OWL (Ontology Web Language) constructions.
The problem of interoperability comes, one might argue, from the absence of universality and from the absence of flexibility within structure that is given to RDF type construction, including the Cyc Corp ontology. (One can make the observation that Cyc Corp has many of the same set of problems that does OWL.)
Comment on this by John Sowa:
http://www.bcngroup.org/Macrocognition/four.htm
with additional comment by myself.
We note that data schema interoperability can occur by restricting the processes that RDF models, and thus leaves out those aspects of natural reality that seem problematic. The part of the problem space we solve gets smaller and smaller.
I developed a presentation for Susan Turnbull’s monthly meeting
http://www.bcngroup.org/admin/technologyReviews/InformationalLatency.htm
and yet each time Brand Niemann talks to me about a presentation at the “next” session, there is simply no follow up. I do not take this personal, of course; but we do regard this as characteristic of semantic web type discussions regarding the relationship between RDF type systems and Topic Map type systems.
One sees the Topic Maps having spectacular success with users in the IRS pilot and yet no dissipation of the Topic Map systems. As every one knows, I have spoken to this issue in many forums. But beyond Topic Maps is a even more simple and less encumbered technology based on precise measurement of structure and the separation of semantics form computer encoding.
Human-centric Information Processing (HIP) is illustrated in my work on subject matter indicators and semiotic systems. HIP builds on a different set of scholarly literatures that does the massive amount of work done in artificial intelligence and related literatures. The work remains under funded, marginalized and even demonized; in spite of the clear need for a fresh alternative.
My work is but the tip of an iceberg of unfunded work by hundreds of innovators worldwide. And yet we have evidence of a pervasive and systemic avoidance of funding that might show the massive funding of AI and RDF work as being less than as advertised. There is no business alternative, and yet the principled alternative is clear from a review of the literature. It does not have to be this way.
A question might be considered in light of two claims:
1) Human-centric Information Production (HIP) shifts the expectation of inference from a deductive chain, a sequence of mechanical steps that must all occur in a computer, to a principled reliance on human in the loop cognitive priming and cognitive acuity.
2) A national educational project is needed to break the common thinking away from the artificial intelligence mythology and break down the will to continue along a path to a Semantic Web where of the two sides (the machine side and the human side) the machine side if far more important.
We address this question to the W3C and to the great thinkers at TopQuadrant.
Do you see any actual
evolution in the proven capability of RDF and FOL (First Order Predicate Logic)
to address the issue of universality over the past five years?
Our judgment is that less has been accomplished on the core issue of universality while a standardization process is forcing unworkable solutions onto those who have intractable problems due to the absence of a single clear standard.
We appeal to TopQuadrant to re-focus its business model around the proposition that the government IT sector needs to look more to the Topic Maps standard and less to the RDF model. We also ask for the community to consider supporting the exploration of an alternative to RDF and Cyc type ontology constructions.
Paul S. Prueitt
Founder, BCNGroup.org; Ontologystream.com