Am 28.09.2015 um 16:41 schrieb Peter F. Patel-Schneider:
I agree that finding the right thing to use is not easy.
However, I think that a uniform search space is better than a non-uniform one. I would much prefer to look through a collection of properties than a collection of properties and qualifiers. If I am writing a tool to help the process, I would much prefer to display a collection of properties than a collection of properties plus qualifiers.
I understand. But you buy flatness of the search space by only having one dimension for modelling, instead of the multiple degrees of freedom you have with qualifiers.
Similarly: of course it would be nice to build an ontology of everythign in the world using a single class hierarchy (taxonomy). Nice and clean and easy to handle. But this approach runs into problems as soon as you try to model a non-trivial domain. The result is usually a very awkward modelling of the domain.
Using a more expressive model (e.g. adding interfaces as in typical OO languages, or adding mixins, facesses, traits, etc) makes the model as such more complex, but the mapping to the domain (in our case: the world) less complex, and more natural.