Sunday, February 26, 2006
Generative Methodology Glass Bead Games
On the limits of the OWL standard ŕ 
Reading material 
Reading material 
Reading material 
Summary of the discussion up to this point ŕ 
The issue of n-articulated ontology
Related discussion on ontologyMapping thread ŕ 
Irene’s comment on the same subject ŕ 
What a nice note. Thank you.
I would like to add to topics to what we have already pointed out as relevant. The need to get quicker progress is not merely something that concerns research communities. The world is in great need for a clarity that can come if and only if the vision of the Semantic Web is corrected and a new vision substituted.
http://www.bcngroup.org/beadgames/anticipatoryWeb/1.htm (for example)
For example, Bird Flu preparation, avoidance (?), and response is differentially impacted based on the rapid development of medical ontology, and reaction ontologies. And would it not be very nice if the "New War" would come to an end, because of both transparency and reaction capabilities - and perhaps the next US Presidential elections. (I am only suggesting that a fresh start is needed.)
The issue of formal verse natural systems has been explored in the 17 pages (so far) of the mapping ontology thread of this "glass bead game" (named after the concept as presented in Herman Hesse's internationally read book), and in the generative methodology thread
And in a chapter I wrote
The second topic has to do with the nature of reality, seen as a "expression"; be it gene, cell, person, social or otherwise.
John Sowa has been involved in this discussions, although I do not see the details of what he proposes as a solution to interoperability between ontological modeling and other types of information models (UML, Object Oriented, RDBMS), I sometimes do not know if he is actually suggesting a solution with the common logic constructions. I am strong critical of the notion that a lattice of theories is relevant to anything, because I (as a pure mathematician) do not see the formal order that will allow the placement of graph nodes (corresponding to "arbitrary" theories) into a lattice.
What I do see is that the assertions of OWL are far too complicated and do not allow certain types of process models, for example the evolution of natural categories.
The real natural ontological expression may not be deterministic. A "true" open world assumption may explain that multi-interpretations of the "same thing" in human communication are in fact an essential part of the real world (ie real nature).
Any ontology that is not n-articulated cannot model human knowledge - this is an opinion. To be n-articulated the formation of the assertions (all of them) have to be formative in real time. It be will be that there are "universals" but these are not likely the class - subclass assertions that come from the current standards.
It is our feeling that
cannot be anything more than an academic exercise. Not only is it not comprehensible to almost everyone, but it does not address real issues that the authors might be aware of if they were not so purely academic.
You seem to agree
I must confess that our group came to the conclusion in 2003 that a linear metamodeling stack is itself insufficient to handle the modeling of actual systems and communities,
The notion of federating information models might be the step that comes next, but OWL might be be interoperable with anything but itself. Many are looking at this question.
In reading your Roadmap I am pleased to the extent that you move away from misleading language like "semantic web". This does not good to hype things in a way that becomes incorrect when viewed from natural science. Meaning is only partially represented as the categories of classification - even if these classification were perfect (and we are both agreeing that the current SW stack is not correct.) Some aspects of meaning (semantics) is situational.
I see also that you are suggesting that UML (ie Object Oriented data models) should have a "translation" engine that expresses OWL, and that OWL should export as a data model.
(page 7 / 110) of your road map.
On page 12 you have a figure that suggests that you feel the Chu spaces and things like Galois theory will be sufficient.
The Second School suggests that these formal system are just big equations, and are tempting because must to not have the background to understand their limitations.
The human being "creates" a sign system through "induction" and this involves more than can be formalized.... at least this is our position.
This can be made to work using the BCNGroup Roadmap
I will continue to read your Roadmap, realizing that a great deal of excellent work is represented here.