Upper and domain ontologies

Hi all:

I have the following doubt involving upper ontologies and domain 
ontologies. Upper ontologies are often described using FOL and modal 
logics because they deal with general semantics and intended meaning. 
Its implementation in the Semantic Web is in DL, much less expressive. 
For example, DOLCE UltraLite is a implementation in DL of the upper 
ontology DOLCE. Therefore, if I develop a domain ontology, which one of 
the following approaches is the proper choice?

(1) Develop the domain ontology in DL, importing a well-known DL 
implementation of the upper ontology.

(2) Develop the domain ontology in DL, importing the upper FOL ontology. 
Then, create a DL implementation importing upper concepts from a 
well-known DL implementation.

IMHO, the second approach guarantees that the concepts of the domain 
ontology are consistent with the intended meaning of the upper ontology. 
However, available tools, such as Protege, are focused in the first 
approach. I think these tools makes easy to develop domain ontologies 
that claim to be based in the upper ontology but can violate some of its 
axioms that can not be expressed in DL.

This question is relevant in my field (GIS). I realize that we are using 
more and more the layered approach to develop domain ontologies, and I 
have the impression that we are just cherry picking the upper ontologies 
(the approach 1) rather than develop something with a shared semantics 
(the approach 2).

Many thanks for considering my questions.



Francisco J. Lopez-Pellicer
IAAA, Universidad de Zaragoza, Spain

Received on Sunday, 29 November 2009 23:10:31 UTC