Hi Markus,
Thanks for your responses. Markus, I think the point that Sebastian was raising has more to do with practices for communities working on data modeling for wikidata than specifically about OWL semantics. Let me explain a little further. We are a group of 3-7 (depending on the week) people working collaboratively on the task of loading wikidata with content linking genes, diseases, and drugs. Even amongst this small group, we have struggled to keep our data modeling discussions orderly and productive - even before entering into these discussions with the broader community. Its a constant struggle to see the big picture. One of the main reasons for this (IMHO) is the lack of ways to view the structure of the model that we are assembling as its being figured out. This is a consequence of wikidata's schema-free design. e.g. on Freebase this problem was addressed using their Type system. For a given kind of thing, you could create/find a Type to describe it and there you could argue about what set of properties were most useful for representing things of that Type. Wikidata seems to want to deal with things one property at a time - which is fine until you want to come up with a coherent collection of a number of related properties and associated constraints that cover a particular domain. For that purpose an ontology and tools for looking at and thinking about the ontology become very useful. So.. currently we are experimenting with webprotege as a place to collaboratively work through our data models before entering into discussions on wikidata itself. Thoughts on that as a pattern for collaboration would be helpful - could/should we be doing this all in wikidata? Would some interface improvements be possible that facilitated schema-level views and discussions?
The idea of working in OWL (though note that we are not currently using any semantics beyond RDF-S) provides the added potential bonuses of facilitating import/export and mappings to other linked data sources, but this is really secondary to the social management challenge.
Emw,
We have not explicitly attempted to force alignment with BFO or OBO - though we have been in touch with Chris Mungall about this and would welcome help with such alignments either on webprotege or on wiki. We are driven very pragmatically based on the requirements generated by the data sources that are next on the list for import but, as the ontology discussion should indicate, want to do our best to help generate a clean and effective model for the community to build upon.
-Ben