Saturday, January 21, 2006
[361] ß parallel discussion in “national debate” bead thread
The Second School of Semantic Science, on
founding
On the limitations of Knowledge Engineering Logics
Professor Chris Menzel
Texas A&M University
And the ONTAC e-forum
I appreciate your clarity in presenting your thoughts about type, set, and class distinctions.
http://colab.cim3.net/forum//ontac-forum/2006-01/msg00119.html
I try to understand why something like set theory and common programming logics (like the programming language "Basic") is not simplified into something sufficient for the purpose of organizing and obtaining retrieval from sets of concept representations.
My opinion is that an unreachable goal is stubbornly held on to, that of creating an "intelligent" computer, where by "intelligent" one means having all of the properties of human or living intelligence. So the easier and obtainable goals are not achieved, because these easier goals are not defined.
To illustrate this point I reference a discussion between Andrea Proli and I recently. I respect very much his command of the many logics floating around but his correspondence shows that the problem may come from a type of KE discipline narrowness of mind:
http://www.ontologystream.com/beads/nationalDebate/357.htm
where Andrea gives some good information about the many varieties of definitions of "axiom".
and then
http://www.bcngroup.org/beadgames/generativeMethodology/134.htm
regarding the viewpoint that the natural scientist's may need to use terminology "complex", "inference", "expressive", etc in a different "namespace", than how this terminology is used by the KE.
If you have read any of my work you see that there is substance.
But I can find no way to get standing for this type of semi-formal methods within the OWL / RDF communities. I do think that KIF was a better foundation then RDF. The reason is that the frame has slots and fillers in a way that allows a localization of context (in the frame), and that a theory of filler type can be developed in accordance with observed phenomenon. Little by way of logic (inference) is required to make this work. One needs set theory only. One also needs a type of category theory that is merely involved with creating symbols for observations.
The formation of categories is where one finds inference, but this type of inference is inductive, the formation of meaning through perception (by a human). It is not deductive, as we find in OWL.
Ideally one would be able to make the distinction between deductive logics related to retrieval of information in an OWL information base, and the formation of ontological structure through induction by human beings.
Well, the note is really just an acknowledgement that you seem to have taken the time to express clarity when others have not.
How might it be that the folks who now are vested in ontologies with logics might be kind to the rest of use and lay out some objective something that is understandable and has closure?
This means, critically, that the knowledge engineering has to be transformed into a knowledge technology that is (common) human centric. This transformation dis-invests in the discipline of KE in order that knowledge technology become more commonly available and usable.
I find hope in the Semantic Web Services Initiative
http://www.swsi.org/ as well as in the new OASIS WSML (Web Services Modeling Language).
The other area I am looking at is SOA IM (Service Oriented Architecture Information Model).
Paul
Prueitt PhD (mathematics and quantum cognitive neuroscience)
The Taos Research Institute
Taos New Mexico