[215] home [217]


Friday, March 03, 2006


Challenge problem

The Taos Discussion


Generative Methodology Glass Bead Games


On the limits of the OWL standard [184]

Reading material [1]

Reading material [2]

Reading material [3]

Summary of the discussion up to this point [186]



On ontological modeling of expression


Subtitle: incomplete information



On Formal verses Natural systems [206]




I will attempt to say something here that might not be very clear, but involves the fact that in natural language most of the information is "not there" and yet this is often not an impediment to the communication.


(Other things like "mindset" can be a greater impediment even when the language is very complete and clear.)


It is simple to say that this ability to have incomplete information, and still actually understand, is due to the filling out property that our consciousness is able to achieve. The "signal induction" occurs, even without all of the (structural?) information.


Text understanding technology has this problem, and for many in this field it seems that only a large sample of text will create a full profile of the "concept" being expressed. I disagree, but that is a longer story.


The concept of Schank's notion of a frame comes to mind. The frame invokes the filling of the remaining slots.


And the term "reification" also comes to mind. For me, reification should be when some process inspects (measures) an abstraction and in the course of this inspection modifies the abstraction in a way that the abstraction should have a higher fidelity to the object under investigation. Reification should involve human curation of a particular ontology.


I am not suggesting solutions to the issue, I could but this is not what I do want to do here... because none of my solutions are complete (and easily understandable outside my head). So I fail to communicate, and perhaps have it wrong anyway. Listening is better, at this point.


Frank said several things recently that caught my eye


"I think the point I am trying to make is that what we call pathways to the largest extent are artifacts created by humans to reduce complexity and be able to discuss complex phenomena."


This process of reducing the complexity is in fact similar, I conjecture, to this issue of merging two ontologies where one ontology has additional information (using the partial open world assumption of OWL) that the other ontology left out.


I feel that the Rosen definition of complexity is properly used in this discussion as a means of unfolding information. But how that might be automated is clearly a discussion for a circle of dedicated scholars working with real data about real projects such as the BioPAX project.


It is an exciting time.

[1] http://dip.semanticweb.org/documents/ECIS2005-A-Methodology-for-Deriving-OWL-Ontologies-from-Products-and-Services-Categorization.pdf

[2] http://www.mindswap.org/2005/OWLWorkshop/sub1.pdf

[3] http://bip.cnrs-mrs.fr/bip10/rosen.pdf