tombaker
commented
3 years ago @dr-shorthair Thank you for raising this interesting issue!
Some history: The DCMI Type Vocabulary goes back to a 1999 proposal from Rebecca Guenther that is still available on the LC website. It is interesting to note that, as originally defined in the first draft, a physical object was "a non-human object or substance. This category includes objects that do not fit into any of the other categories on this list". As originally proposed, "DC-Type" was a "qualifier" for qualifying the Type element. Prior to the use of URIs as term identifiers, it was intended as a keyword for giving context to a text value such as the string "physical object".
In 2000, it was approved as an Element Encoding Scheme but eventually evolved into the Vocabulary Encoding Scheme DCMIType and defined as "the set of classes specified by the DCMI Type Vocabulary". (One can think of the URI of a Vocabulary Encoding Scheme as being like the concept scheme URI of a SKOS concept scheme.)
Looking at DCMI Metadata Terms today, one might question whether the classes of the DCMI Type Vocabulary are substantially different from the classes defined in the /terms/ namespace. Classes such as "Agent" and "LicenseDocument" might just as well be used as the value of dcterms:type (or indeed rdf:type) assertions as "Collection" or "Dataset".
It is in this spirit that I point out the definition of dcterms:PhysicalResource: "A material thing". I'd argue that an organism is a material thing.
I'm curious whether this would fix the DWC issue or, if not, whether it were because users see a significant distinction between classes defined in /terms/ as opposed to /dcmitype/.
Current definition:
This would appear to limit the applications of DC. What DC Type should be used for descriptions of living things?
(Asking for my friends over in the Darwin Core community, who are trying to clean up the definition for samples and specimens.)