Thanks for the extensive comments, Roman. I agree: these days, luckily, UML is imposed upon the engineers as just a few years ago (depends on the industry).

Now, to ontologies. (Please excuse me if I am sloppy below.) Ontologies make sense and are useful precisely for the reasons you described.

Two things, though: they are symbolic systems (see my previous post) and they may be expensive.

Firstly, since they are symbolic systems (almost by definition), they have their place in the software design only as long as their limitations are kept in mind: symbolic systems cannot express things that semiotic systems can (programming languages are semiotic systems), which seriously constrains their usage. Ontologies are not languages (there are languages to express ontologies).

(By the way, I am not sure why you referred to description logic, but it is a good example: description logic is less expressive than first order logic. Brandom’s book I cited (Between Saying and Doing) offers a formal proof of expressive limitations of various logics and shows how they can be overcome by introducing semantic-pragmatic (meaning-use) relationships.)

Secondly (briefly), ontologies are “expensive”: they are ready-made hierarchies, which one would need to design or reuse: they may move quickly towards the Laplace’s daemon situation: lots of software design situations are very specific and circumstantial and one would approach them first not from the point of view of ontology, but from user stories prospective, trying to design a mini-language (as small as possible) that both the user/client and software engineer could relate to.

I wish I had time right now to comment more extensively, hope we talk more in future.