On Sep 21, 2011, at 5:11 PM, Gregor Hagedorn wrote:
- generic part-of relations of concepts.
Example: the taxpub nomenclature section is part of the taxpub treatment We consider using DublinCore for this. We found nothing in SKOS for this.
I would use DC partof / haspart for this. This, I think, will make you most likely to use the same vocabulary and properties as many others who state relationships between publications, their parts, and things connected to them (such as, for example in the case of Dryad, data packages and their parts).
- part of relations of things: both petal and sepal are part of the
corolla.
I would use RO. This is a well-defined relationship, and will make you use the same as many others who express relationships between anatomical parts, or organisms and their parts.
Bob Morris recommends the ro ontology: http://obofoundry.org/ro/ However, this one says that it is undergoing strong changes in the near future.
I wouldn't worry about this (but see below). The plan is to roll (most of) the relationships currently in RO into BFO (Basic Formal Ontology). They'll continue to exist, but will then receive different URIs. Since there are lots and lots of ontologies and projects that use RO (for example, using it is one of the OBO Foundry criteria for membership), there's already a requirement to make this reasonably painless for users, which is perhaps part of the reason that it still hasn't happened yet.
The one caveat is that once these are in BFO and you apply them as BFO properties, by doing so you subscribe to the BFO worldview for semantic purposes, strictly speaking. For most practical purposes and applications that's likely of negligible if any consequences, except if you plan to make processes (in the BFO sense) bear qualities (in the BFO sense). And there's something about 3D immaterial extents in BFO that's the subject of controversy, but I forget the details (which says something already).
- When documenting legacy xml ontologies, the xml variant requires
documenting a difference between element and attribute children.
What's an XML ontology? Isn't that an oxymoron?
-hilmar