[tdwg-tag] Darwin Core term management - dc:type, basisOfRecord, and recordClass

[John R. WIECZOREK]

A fair bit of the confusion no doubt arises from the Decisions example
for Behavior. Allow me to clear that one up first. That decision was
meant as an example based on something that really did happen during
DwC development. It's only one of hundreds of decisions that were made
during development, and isn't meant to persist once we have a real
Decision to put in place. At the time that decision was made and
placed in that page, "Behavior" really was the term name, not
"behavior" - that change came later. Behavior really did have a domain
(rdfs:domain) of Event at that point. Subsequently all domain
assignments were removed for reasons given elsewhere, but summarized
as DwC is not meant to be an ontology, but rather a library of terms.
The alternative to the formal domain was a new attribute
dwcattributes:organizedInClass, and as you surmised, this is where we
make suggestions about which Class (a term having <rdf:type
rdf:resource="http://www.w3.org/2000/01/rdf-schema#Class"/>) a
property term (a term having <rdf:type
rdf:resource="http://www.w3.org/1999/02/22-rdf-syntax-ns#Property"/>)
should be associated with for human consumption.

A. So yes, you got it right, the object of the predicate
dwcattributes:organizedInClass for dwc:behavior is now meant to be
http://rs.tdwg.org/dwc/terms/Occurrence.

B. Yes, you got it right again. The URI given for the Class in the
documentation (see
file:///C:/Data/DwC/darwincore/terms/index.htm#behavior)  means that
the object of the predicate dwcattributes:organizedInClass is the
stated Class.

C. Correct. Such an implication could end up faulty in an ontology
built from these terms. The term dwc:scientificName, for example,
could be used in different contexts.

D. Correct. The rdf:ID attributes will be removed and the white space
problem fixed in the next commit to the code site, with like commit to
rs.tdwg.org.

1. There are currently no domain assignments for terms in DwC. If a
term has a URI next to it in the documentation (Quick Reference or
Complete History), the URI is determined by the predicate
dwcattributes:organizedInClass. If it has the literal string "all" in
these pages it is meant to be a "record-level" term, which means it
could equally be organized in any of the DwC classes and would apply
to a whole record, regardless of the dcterms:type, dwc:basisOfRecord,
or proposed dwc:recordClass.

2. Correct. No DwC terms has a predefined range. Only the dcterms do.

3. "Domain" always means domain in the formal sense of rdfw:domain. It
is fair to say, as you did, that "an extension of DwC to valid OWL
should not assume that any relation between rdfs:domain and
dwcattributes:organizedInClass is intended by the DwC, which is
entirely silent on both rdfs:domain and rdfs:range for all of its
properties."

[I think your distaste for the interpretation above may be tempered by
the fact that the cited change to behavior occurred while domains were
still considered. They have since been abandoned to avoid
inappropriate assertions, such as a resource having a property
dwc:scientificName must be an instance of a dwc:Taxon. Such assertions
are to be made in ontologies, and distinct ontologies may differ as to
the domain of DwC terms. The idea is to build simple ontologies for
specific world views, and to construct world views of larger scope by
constructing further ontologies from these for particular purposes.
DwC makes no claim to understand or enforce any given world view.]

2009/11/18 Bob Morris <morris.bob at gmail.com>:
> I am confused about the usage in DwC documents of the phrases
> "domain",  "Class", and dwcattributes:organizedInClass
>
> A. The file http://rs.tdwg.org/dwc/terms/history/decisions/index.htm
> carries the text
> "The domain of Behavior was recognized as having been spuriously set
> to Event when it should have been Occurrence."
> (This should be "behavior" not "Behavior" I believe).
>
> As best I can tell, this is meant to mean in this case that the object
> of the predicate dwcattributes:organizedInClass is now meant to be
> http://rs.tdwg.org/dwc/terms/Occurrence
>
> B. http://rs.tdwg.org/dwc/terms/index.htm carries the text
>   "3. Term Definitions
>     [...]
>     Class:     The primary category of terms (Class) that this term describes.
>     [...]"
> As best I can tell this means that the object of the predicate
> dwcattributes:organizedInClass is  the stated Class
>
> C. The definition of organizedInClass in
> http://www.tdwg.org/standards/450/files/darwincore/terms/attributes/dwcattributes.rdf
> carries  the rdfs:comment
> "Does not imply domain or range membership in the class."
>
> D. (Irrelevant to my confusion, but needs to be fixed):
> dwcattributes.rdf is
> --- ill-formed XML on trivial grounds (there is white space after a "<")
> ----  is invalid RDF even when made well-formed because it has both
> "ID" and "about" attributes on all four dwcattributes that it defines
>
> So below are my conclusions. Are they right?
>
> 1. "The Class of a dwc property p is X" and
>  "The domain of dwc property p is X" both mean
> p  dwcattributes:organizedInClass  X
>
> 2. "The range of dwc property p is X" has no expression in the current
> dwcattributes or elsewhere in DwC
>
> 3. The two informal uses of the word "domain"---one in the decision
> file cited and perhaps elsewhere, and one in the rdfs:comment cited
> from dwcattributes.rdf ---are inconsistent, but the latter may be
> interpreted as: "an extension of DwC to valid OWL should not assume
> that any relation between rdfs:domain and
> dwcattributes:organizedInClass is intended by the DwC, which is
> entirely silent on both rdfs:domain and rdfs:range for all of its
> properties."
>
> [Personally, I don't like the above interpretation. The very fact of
> having made the cited change to "behavior" to me suggests that DwC
> does care about informal semantics, and that at the very least,
> organizedInClass is informally intended to be a subProperty of
> rdfs:domain. ]
>
>
>
>
>
>
>
>
>
>
>
> --
> Robert A. Morris
> Professor of Computer Science (nominally retired)
> UMASS-Boston
> 100 Morrissey Blvd
> Boston, MA 02125-3390
> Associate, Harvard University Herbaria
> email: ram at cs.umb.edu
> web: http://bdei.cs.umb.edu/
> web: http://etaxonomy.org/FilteredPush
> http://www.cs.umb.edu/~ram
> phone (+1)617 287 6466
>