FOIS '10 -- Ontological Diversity, Not A Tower Of Babel

A celebration of diversity is not what one would expect from what a colleague characterized as the "dry world of ontology".  Yet this was the topic of several presenters at today's morning sessions at FOIS 2010.  Especially, at this morning's "Invited Talk", by John Bateman, Professor of Applied Linguistics at the University of Bremen, the topic of ontological diversity was presented as both requirement and artifact for successful ontological engineering.

Perhaps the reader should not be surprised at the juxtaposition of diversity and ontology.  If ontological engineering is about modelling the diverse richness of reality, then intuitively it only makes sense that ontologies themselves should also be diverse. 

From a business perspective, Dr. Bateman underscored a theme from the conference, which is that (interpreted in the words of your host) "a one size fits all ontology" doesn't work and further that "a rhetorical demand for ontological simplicity" doesn't work either. Reality is rich, reality is diverse, and only high-fidelity ontological tools will make the desired impact on a new generation of software.  Offline, Dr. Bateman shared examples from his work in spatial services associated with healthcare.  The practical and important nature of this work is a small proof of the positive impact that can be made by good ontological engineering.

Dr. Bateman suggested that a successfully ontology-based system would be characterized by deployment of multiple ontologies, what might be called a diversity of ontologies, each one serving a specific purpose.  Dr. Bateman made the point that there is no common upper ontology that can be expected to tie together each domain-specific ontology.  Rather Dr. Bateman proposed a "hyper-ontological" approach, which seemed to be almost a "federation" of ontologies.

It is conceivable that a "super-ontology" could capture all the diverse richness of some target reality.  Why does Dr. Bateman then recommend that successful ontologies must also be granular?  It seems from experience that monster ontologies are usually functional failures.  Such over-reaching ontologies are both error-prone and unmanageable [ -- and therefore probably useless].  Your host adds that if one adds in the inevitable, constant growth or further refinement of a good ontology, required to match an evolving understanding of reality, the management and error problems associated with the super-ontology will only be exacerbated.

Readers who are interested in Dr. Bateman's comments can refer to "Ontological diversity: the case from space", John A. BATEMAN, Formal Ontology in Information Systems, A Galton and R. Mizoguchi (Eds.), IOS Press, 2010.  Any corrections on the notes here, which are solely the responsibility of your host, are welcome.

Draft Version 0.4