August 10, 2005
Center of Excellence Proposal
à
ONTAC stands for Ontology and Taxonomy Coordinating Working Group
It is a working group of
Semantic Interoperability Community of Practice (SICoP)
Communication from Chris Catton to Patrick Cassidy (two regulars on the ONTAC e-forum).
Pat,
I think we seem to be moving towards an agreement that a *very* simple
upper level ontology that captured the broadest consensus of the community
about basic ontological relationships would be useful. Wide use of this *could* be encouraged by
the W3C best practices group, and it *could* be made available as a default
import into OWL-protege.
Comments I made and posted to the e-forum:
The W3C community generally shares some underlying assumptions:
1) that individual terms should have well defined and specific meanings
2) a classical logic should be available that uses these terms as if part of a mathematical framework
If one has both of these things, well defined meanings and a logic with which to compute truth and determine falseness, then we will have “machines that know each other” and “machines that reason about everyday affairs as if a super intelligence”.
The notion that there should be the possibility of logical contradiction with "ontology" appears to many non-computer scientists to fly in the face of a reality. The reality is, however, one related to the world as the W3C would have it be.
Logical consistency is a hard thing to define when one is using computer representations of human concepts.
The W3C community treats the natural world as if it is reducible to a logical representation.
The alternative is to use the co-occurrence relationship and to develop machine ontology that is about the structure of co-occurrence relationship with the meaning of these structures left to be interpreted by knowledge humans, see [*] and [^].
Natural ontology is not logically consistent, has not been found to be logically consistent. This is the fact that is ignored by most computer scientists, in spite of Godel's work and the work of others (Wittgenstein, Peirce,
http://plato.stanford.edu/entries/peirce
Penrose, Rosen, and others).
Phase coherence is a local phenomenon, and from this biology has found the support for cognition (Pribram).
The comment that a simple upper fixed ontology might be encouraged by the W3C best practices group is a most interesting one, one that most in the W3C community have supported for a long time.
But is this suggestion truly informed by an awareness of the consequences of fixing meaning of terms like "attack", or "friendly"? What about "is a"? Must every relationship be an "is a" relationship? Most natural scientists do not think so ! Should the military use a simple fixed upper ontology with terms like "military" and "civilian"?
Even if one is talking about a
small simple upper ontology for all circumstances, the fundamental assumption
(one made with great authority by the W3C) is that one-word one-meaning
one-location-in-the-Semantic Web is the ideal.
The alternative is reconciliation containers were those terms that are used in normal discourse have properties indicating multiple possible meanings and some contextual information. Even then there is the need to push away the polemics that suggest logical systems and fixed meaning is better than the complex expression seen from the every day human use of natural language.
http://www.bcngroup.org/area2/KSF/nationalProject.htm
A Roadmap for the adoption of semantic science is given at:
http://www.bcngroup.org/area1/2005beads/GIF/RoadMap.htm
Dr Paul S Prueitt