When a UML diagram or XMI export is generated with OntoUML stereotypes enabled, object types that were modelled top-down as concepts receive no stereotype at all, while types derived from fact types are classified normally.

This is the wrong way round: the top-down layer is exactly where the modeller is describing things directly, and where a classification such as kind, category or collective is most meaningful. Many of these types already carry a concept type that could drive the choice.

Expected: concepts are classified like other object types, falling back to a sensible default instead of being left blank.