A couple weeks ago I gave a talk at RPI on ontology design I called "Careful with that Predicate, Eugene". For some of us, this might seem to be some straightforward ontology design principles, but for others, it might be either new or an approach that they actively dislike. While I don't want to blow up the mailing list, it seems like this might be a good place to discuss ontology design among actual practitioners.
I'm happy to receive feedback both positive and negative, as I contemplating turning this into a more formal ontology design paper. It's about an hour long. If you're new to ontology design it might be well worth your while.
Ontology design methods can vary from creating lots of properties (like in schema.org, Wikidata, or ontologies designed from UML) to few properties, as in Semanticscience Integrated Ontology, Basic Formal Ontology, and PROV-O. We will explore the consequences of these design decisions and try to establish some general guidelines for when a new property should be introduced into a data model, ontology, or knowledge graph, and when general purpose properties might be more useful.