Hi Steve,
I really need to read your other messages before I can commpent in full, but I did want to respond to some of your recent comments, below.
What if the specimen is from a tree? I collect flowers in May and return
to
collect fruit in September. I have a hard time making those two Occurrences be one.
Why would the two occurrences have to be one? You have an Individual instance for the tree, and a two more Individual instance for the collected subsample. The Individual instance corresponding to the tree could have its own related Occurrences (e.g., if you took a photo of the tree); and the two collected specimens would each have their respective collecting event represented as an Occurrence. The latter two occurrences can trace back to the Individual Occurrence by virtue of the fact that both of the collected Individuals are semantically linked to the "parent" Individual (i.e., the whole tree). Isn't that the sort of "reasoning" that RDF is supposed to be able to allow? That is:
Tree exists as LivingSpecimen Individual whole tree
Flowers exist as PreservedSpecimen Individual that is part of a tree
Fruit exist as PreservedSpecimen Individual that is part of a tree
The first can have as many Occurrence linked to it as needed (images, measurements, etc. all taken on different dates).
The latter two likely have one Occurrence each (the collecting Event for each).
The latter two Individuals are linked as "derivedFrom" or "partOf" or whatever back to the whole Tree Individual, so that the presence of the whole tree can be inferred by the Occurrence records for the two parts (specimens).
According to the definition on the table for a vote, an Individual is to permit resampling over time.
Absolutely! No argument there! You can go back to the WholeTree Individual and generate a zillion Occurrence reocrds if you want. Not a problem.
You are denormalizing a more general model. We both confessed to this sin
in an earlier series of emails.
I don't understand what you mean by that.
If you don't want to read the first post in the series, just click on the links on order and look at what happens to the diagram. I want (no NEED) the fully normalized model for what I do and so do others.
And so do I.
You may not need it,
No, I do.
If I'm the only person who ever needs to resample anything
You are not. Nothing in my proposal prevents resampling of an Individual. In fact, nothing in my proposal prevents *anything* you want to do, as far as I can tell. But your proposal prevents me from representing parts of an whole organism as Individuals unto themselves, which is especially encumbering when I know that I have many, many, many parts of whole organisms for which other parst of the same whole organism exist, but I lack the knowledge to build those links.
I would just prefer it to be a "well known" Darwin Core term, not an ad hoc one that I made up.
I would too! But in the same way that DwC started focused on DeadSpecimensInMuseums only, and was later expanded to a more general accomodation of occurrence records for both dead and live things; your proposal for "Individual" may have started out to fulfill a very specific need, and I am advocating a slightly broader interpretation that still meets your specific needs completely, but also accomodates a borader scope of needs (without denormalizing anything).
Maybe we need a new term like "BiologicalObject" that should be the more general class, of which "Individual" as a more narroly-defined subclass?
If it was important enough to put individualID into the Darwin Core standard to facilitate resampling, then why is it suddenly not very important to make that term usable in RDF?
I can't speak to how it would be used in RDF (I am Bob's "BETA"); but I can see how the same definition of "Individual" can accommodate both sets of needs rather elegantly. More after I digest your other two messages.
Rich