Monday, November 21, 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 to the working group
from Paul Prueitt
****
(Additional comments on the nature of formalization using logic) à [229]
First, a practical problem: Ontology editors, in particular Protégé, have not become easy to use or stable. A tremendous amount of detailed knowledge of J2EE, Java, the specific of the Protégé code base (and continual updates) is necessary. Domain knowledge is also necessary. Without having a simpler editor, those with domain knowledge cannot participate.
Having said that, there are many new applications being made using Protégé and Protégé interfaces. Solutions to foundational issues, such as those addressed by this ONTAC working group, may lead to a stable editor someday. The issue addressed by many has to do with the RDF and OWL standards themselves, since these standards deal more with technical issues of programming in ways to may not be compatible to new understanding of the foundations of ontological modeling.
Most in the working group are in agreement that there are unresolved foundational issues. These foundational issues are addressed differently across that community working to specify ontology structure. Foundational issues are addressed, also differentially, across the community working to build software that edits and uses ontology. Over formalization requires that the software community dictate to the domain knowledge community. Is this a problem?
Second, a simple solution: Ontology can be defined as an enumeration of concepts and relationship between concepts. This means of course that the word is double articulated, with two meanings
1) ontology is reality
2) ontology is “about” our concept about reality
One way to disambiguate this dual meaning is to use the phrase “ontological model” and “ontology”.
In this disambiguation, the term “ontology” is improperly used when used to refer to a model equipped with inference logic.
The distinction made is the same as the one where Hilbert mathematics models of neural networks should be called mathematical models of neural networks and not “neural networks”.
This distinction between the formalism and the reality might seem trivial and of no interest; however the literature in neural networks, for example, is rich with examples where properties of a biological system is attributed to a computer program that is implementing a model of that biological system. In many cases, this attribution is not justified and yet creates its own “truth” (it was said in an article by so and so that the “neural network” performs an understanding of the patterns of data presented to the computer program. ) The literature assists in the propagation of something that is not true about the computer program. The means for this propagation is the confused use of the term “neural network”.
**** Syntax, semantics and pragmatics
A taxonomy or controlled vocabulary can be equipped with some structure.
This structure can be linked with what are called “reasoning engines”. But the fact is that taxonomy and vocabulary can be stated, agreed on by a community, and used without excessive imposing of “is-a” and parts-of” relationships. The absence of ontological statements, “this is a part of that”; can allow the meaning of the terms to shift in ways that are flexible and indicative of pragmatics.
Associations and attributes are also sometimes treated as ontological statements rather than formal statements (used as a convenience). The BFO relationships
http://genomebiology.com/2005/6/5/R46
are an example of a set of relationships that are a good attempt at providing formalism.