Once upon a time there was a knowledge engineer called KE. He was given two tasks: 1. to construct a taxonomy of all the existing mammals on earth, and 2. to construct another taxonomy of all the humanoids in Star Wars. How did he proceed?
In the case of the earthly mammals, KE started to read relevant zoological literature as well as to interview taxonomically interested zoologists, i.e., the domain experts at hand. Since KE was a very good knowledge engineer, he very quickly understood the contemporary scientific mammal classification and turned it into a good computer system. In the other case, KE started immediately to see the Star War movies, read about them, and to check on internet what the domain experts now at hand, the real Star War fans, had to say. Immediately, he found on internet a list of the humanoids that occur in Star Wars. The Abyssin species was said to be “a swarthy-faced humanoid with a single slit-pupilled eye, a thirsty representative of the cyclopean species”; the Aqualish species was “a bellicose humanoid species with some superficial properties both aquatic and arachnid; they have hair-lined faces with large, glassy eyes and mouths dominated with inwardly-pointing tusks”; and so on. After some work, as expected, KE had classified the humanoids into classes and families and put his taxonomy on internet. So far, so good. And no essential difference between the two cases.
However, after a while KE found one indeterminacy in each taxonomy. In both cases, he went through all his material once again, but both the indeterminacies remained. What to do? His reputation as a good knowledge engineer seemed to be at stake. With respect to the fictional humanoids of Star Wars, KE’s problem came to end when he happened to read Roman Ingarden’s The Cognition of the Literary Work of Art (Evanston, Ill.: Northwestern University Press, 1973). He then realized that fictional entities necessarily have some “loci of indeterminacy”. What is not described by the creators of a fictional entity is simply indeterminate. And that was exactly the case with the problematic humanoid. Nothing could and nothing should here be done by a good knowledge engineer.
With respect to the mammals, KE found quite another solution. He asked a zoologist to make some completely new observations related to the indeterminacy at hand. Since the mammals to be classified were really existing species, KE realized that there are in principle an infinite number of possible new observations that can be made. As things turned out, the new observations did in fact solve his indeterminacy problem. KE was quite happy that he had not tried to use the same methodology to solve both his problems.
When a knowledge engineer meets domain experts that are experts in relation to some areas or aspects of the real world, he has, whether he is aware of it or not, a choice. Either: he behaves as if what the experts tell him is all there is to say, that is, he treats them as authors of a story about fictional objects that very well may contain indeterminacies and even inconsistencies. Or: he takes a realist perspective and keeps an eye to the possibility of interaction with the world itself; if the domain experts contradict each other, then he asks them to reconsider their views.
Proposal: Fictions apart, the default position for every ontology creator should be the realist position; concepts are primarily links to the world. In this respect, there is no difference between application ontologies and reference ontologies. 2 Concepts are Tools.
When one is using glasses, microscopes, or telescopes, one sees through them in order to look at something in the world. But something similar is true also of the tactile sense in relation to ordinary tools. When one is real good at using a tool, it disappears from the foreground of consciousness, i.e., one is no longer intentionally directly directed at it. Instead, one is directed at that in the world which the tool affects. For instance, when I write with a pen, I am hardly aware of the pen, only of what I am writing. I do not feel my fingers holding and touching the pen; if I feel anything at all, it is rather the pen pressing against the paper I am