Yes, I agree with statements made by Bob, Kevin, and Rich: if we get the class Individual, we need to think carefully what properties go with it. The only one that is obvious to me at the present is individualRemarks. In line with what Bob said, I think that figuring out how we will use terms and trying to model relationships and properties in RDF will make it more clear what terms can be appropriately used with instances of various classes and whether it "works" to have a term be applied to several classes. I think individualCount is a case in point. At first it seemed to me that it should be in the Individual class, because it's a property of Individual the way I've defined Individual. But since the count may change with time, then maybe it should be a property of Occurrence (which are essentially observations of the Individual at different times). But how often is anyone actually going to track the number of individuals over time? I could see it happening with wolf packs, but in that case one would probably rather track each individual wolf rather than calling the whole pack an "individual". So what makes sense depends on how people need to use individualCount.

Rich commented that he wasn't sure why catalog number wasn't in the Record Level terms. I'm also wondering about that, particularly since I would use a catalog number with a token (e.g. PreservedSpecimen, StillImage) or Individual (if it were a living specimen) but probably not for an Occurrence (although somebody else might). I believe that in a previous discussion of the xxxxxxxID terms on tdwg-content (can't check because Internet is out at my house and I'm working offline) someone said that all of those terms should go into the Record Level terms because one might use them as properties for various classes. But that hasn't happened. Is that because nobody has gotten around to it or because it takes some kind of official action to "change the standard"? To recap that conversation from memory, it was agreed that people appropriately use the xxxxxxxID terms two ways: as column headings in a database table to indicate what the identifier is for the row, and as ID references connecting the row to records in other tables. In the first kind of use, it makes sense to have xyzID in the class xyz, but in the second case it does not. You can see an example of these dual uses in the DwC XML guide. This is another example where some clear instructions on the use of terms in the DwC documentation would help (i.e. to say that both of these kinds of uses are acceptable). When I was trying to understand Paul's email on Taxon and Name, I was having trouble because I was continually confused about which of the two ways of using xxxxxxID terms he was intending as either would apparently be correct.

With regards to "individualScope", I was assuming that could be covered in individualRemarks. I suppose individualScope could have a controlled vocabulary and individualRemarks not. But again, I think we can see how people want to use Individuals before deciding what terms need to go with it (assuming that Individual gets added to DwC).

Steve

Bob Morris wrote:

On Sun, Oct 31, 2010 at 4:21 PM, Kevin Richards <RichardsK@landcareresearch.co.nz> wrote:
> Not sure if this has been mentioned as I have struggled to keep up with this thread, but it sounds to me like the benefit of the Individual class/properties is to be able > to link together various web resources that refer to data obtained from the same individual in some manner, so we probably need terms that allow the description of > how these individuals, or parts of individuals relate to each other. The Scope idea will help, but maybe there is a need for terms like "partOfIndividual",
> "derivedFromIndividual"?

Now you're talkin Kevin! Actually, now you're talking about ontology, and I plead: Go slow, develop use cases; develop competency questions; develop tools. I note that Steve was careful to separate the question of adding terms to the normative, representation free, DwC, from the problem(s) of making an RDF representation of same:

On Sun, Oct 31, 2010 at 10:38 AM, Steve Baskauf <steve.baskauf@vanderbilt.edu> wrote: >
> I am pleased with the significant and thoughtful discussion that has taken place on the tdwg-content email list
> regarding the relationships among Occurrences,
> Individuals, and other entities that are a part of the community's thinking about biodiversity metadata
> and the way that those metadata are structured. [...] I feel that > would be critical for
>facilitating the ultimate development of a recommendation for the representation of Darwin Core as RDF.
> [...]

This separation is important, because what use one intends to make of an RDF representation has a lot of bearing on what "gotcha's" one has to take care about.

For example, if there is a desire to exploit formal semantics available for RDF stack ---which will probably emerge as a requirement once one starts talking about relations between properties---then different surprises will emerge from the pitfalls if one "merely" wishes to put SKOS relations on the properties and reason about the SKOS instead of the science. But I guess, for example, that Miranker's Morphster project [1] will benefit most in its current use cases, from good mereological ontologies for descriptive data, not just stuff like "more general than".

There are surprises even in the simplest use of RDFS and formalisms about classes. I've previously whined about premature assignment of rdfs:domain while conceding (did I???) that it can sometimes make a designer's intention clearer to humans. Perhaps more startling is that type assignment automatically "creates" an rdfs:class if one was not already available, due to the formal semantics of rdf:type [3]. Thus, in an earlier posting, Paul Murray has (unintentionally?) introduced a new class apni:TaxonName in 33407.rdf [2] via
<rdf:type rdf:resource="http://biodiversity.org.au/voc/apni/APNI#TaxonName"/>

Then there is the question of adequate tools for the desired style of ontology architecture. The OWL community's important tools are not friendly even to DublinCore, whose style is(?) what DwC follows. (Steve Baskauf has complained to me in private email that the Manchester validators don't seem to even check rdfs vocabulary correctly; Paul complained that Protege4 makes a big mishmash of Properties when importing DwC. (Both of these are probably false positives in cases of insufficient typing of properties themselves, and the OWL community probably doesn't care about the origin or utility of such weak typing [4]. )

So the hard part is yet to come. But I agree with you. It is quite appealing.

Bob Morris

Robert A. Morris
Emeritus Professor of Computer Science
UMASS-Boston
100 Morrissey Blvd
Boston, MA 02125-3390
Associate, Harvard University Herbaria
email: morris.bob@gmail.com
web: http://bdei.cs.umb.edu/
web: http://etaxonomy.org/mw/FilteredPush
http://www.cs.umb.edu/~ram
phone (+1) 857 222 7992 (mobile)

[1] http://www.cs.utexas.edu/~miranker/studentWeb/MorphsterHomePage.html

[2] http://biodiversity.org.au/apni.name/33407.rdf
[3] http://www.w3.org/TR/rdf-schema/#ch_type

[4] https://mailman.stanford.edu/pipermail/p4-feedback/2009-October/002448.html

--

-- 
Steven J. Baskauf, Ph.D., Senior Lecturer
Vanderbilt University Dept. of Biological Sciences

postal mail address:
VU Station B 351634
Nashville, TN  37235-1634,  U.S.A.

delivery address:
2125 Stevenson Center
1161 21st Ave., S.
Nashville, TN 37235

office: 2128 Stevenson Center
phone: (615) 343-4582,  fax: (615) 343-6707
http://bioimages.vanderbilt.edu