I believe that as ontologies become more mainstream, driven especially by recent publicity received by the semantic web efforts, more and more people are interested in building ontologies. Not surprisingly, some subset of those people are interested in techniques and methodologies for helping them build high-quality ontologies. Our papers lay the basic foundation for developing good ontologies, with a particular focus on how to correctly (and incorrectly) specify and constrain the use of the subClass relation.

There are not many papers around that address the difficult problems of ontology design, specifically those related to the nature of the concepts we use. Our paper is a first attempt at using simple formal constraints in order to make our intuitions clear.

Although primarily the latter—the paper describes a methodology for Ontological Analysis— we did a lot of work looking at existing ontologies and discovering examples of mismodeling. The paper also describes, specifically for the computer science community, the work done in analytical metaphysics, especially in the categories of identity criteria and sortal analysis. Well, in fact, the big discovery is the practical utility philosophical analysis can have for computer science. But the paper is clearly concentrated more on the methodological side.

In AI, database, and now web systems, it is becoming increasingly important to develop a model of some portion of the world that is relevant to the system. The increasing complexity of these models have made it necessary for some to adopt a more rigorous methodology for developing them. OntoClean is the beginning of such a methodology. This paper helps to design better ontologies, by forcing people to reflect on the ontological assumptions underlying the concepts they are currently using, and to be aware of the logical consequences of such assumptions.

Nicola Guarino:
I started getting involved in ontological issues around 1990. I was working on medical expert systems, and I realized that they were totally inadequate for representing the complexity of the medical domain. Also, I was annoyed by the fact that knowledge representation research was more and more focusing on reasoning issues, while the core problems of getting the right representations were not receiving that much attention. I started attending regular meetings at the philosophy department (on philosophy of language), where I met Pierdaniele Giaretta, with whom I wrote some early papers underlying the OntoClean methodology. At the same time I discovered a paper by Winston, Chaffin, and Herrman (published in Cognitive Science) on the problems of modelling parts and wholes, which exposed me to a radically new perspective and also introduced me to the philosophical literature on formal ontology.

Christopher Welty:
I spent a year on sabbatical from Vassar College in Italy working with Nicola Guarino, who had been developing some of the underlying ideas for this research for some time—since at least 1994. Together we were able to work out the formalizations rigorously and, perhaps more importantly, we developed several implications of the framework—in particular the constraints on the subClass relation, the ontology of properties, and the set of common modeling pitfalls. These latter, I believe, made the formal foundation of the paper more obviously relevant to ontology builders and contributed to the paper’s current high citation rate.