The OntoUML classification of types and relationships is only visible in UML output today. Every other artifact generated from the same model - SQL, XSD, JSON, OWL/RDF, SKOS, ERD, documentation - says nothing about whether a type stands for an independent thing, a relationship, an aspect of something else, or an event.

That is the information a receiving system needs in order to understand what a model actually commits to. Two systems can be built on the same model while assuming quite different things about it, and nothing in the exported artifacts reveals the difference. Carrying the classification along makes the assumption explicit wherever the model travels.

Expected: each export states the classification in the way that is native to its format. Where the format has a proper mechanism, use it rather than a comment; where it has none, a comment or annotation is enough. Preferably switchable per export, and only once the classification itself is right (see related tickets).