Comments inline
Paul Murray wrote:
...
How
about:
An
Occurrence is the reification of an individual's involvement in
(entaglement with? presence at? relationship to?) an event. It reifies
an "Event involvesIndividual Individual" fact.
The need for this construct is that we often need to say a
number of additional things about an individual's involvement with
(presence at) an event beyond simply assertin that there is some
relationship. We need to say what tokens that individual left, what
role that individual had (Predator? Prey? Parasite?), perhaps temporal
or other limits of that particular individual at the event. Occurrence
is the object to which these facts may be attached. An individual might
meaningfully have more than one occurrence at an event - particularly
in cases where events are part-of larger events, or where an individual
somehow has multiple roles (hyenas chased away from their kill by a
lion - or is it the other way around?).
To put it another way: "reification" = "tuple" = "association
table" = "pulling a property out into an object". More or less.
Yes and I would add "=RDF graph node" to your list above. It is a
relationship made into something onto which we can slap an identifier
and link other resources via object properties. I would add that its
nature is similar to that of dwc:Event (which you could say is a
reification of an Occurrence's relationship to a Location and time) and
a dwc:Identification (which you could say is a reification of the
relationship between an "Individual" and the Taxon that somebody
asserts that it represents). None of these things (Occurrence, Event,
or Identification) have a physical existence. They exist because we
need them to connect instances of other classes of resources. We give
them names that have something to do with the nature of the connection,
but they really are just connections (or nodes if you wish).
Peter DeVries wrote:
> Wouldn't the individual be asserted to be an instance of a species
"concept"
Defining an Individual this way assumes that one knows what taxon the
Individual represents. What about the Individual I've assigned the
GUID http://bioimages.vanderbilt.edu/ind-baskauf/70858 ? I don't
really have any idea what it is. That really shouldn't prevent me from
assigning it a GUID and an rdf:type. I suppose that you could do what
Rich suggested which was to require at least one dwc:Identification at
a level as high as "Life" but I would prefer to say that an Individual
could have 0 to many dwc:Identifications, with "many" being the number
of Taxa that people wish to assert that it represents. That number is
never fixed because somebody could always come along later and assert
that it was something else, or even assert that it is the same taxon
that somebody else previously had asserted.
Richard Pyle wrote:
>
> I think the hardest part will be to define the allowable scope of
“Individual”. In my mind, it should at least span from single organism
to multiple organisms up to
>colony and population; and can be circumscribed by any taxon
concept (including “Life”). I have been re-thinking whether “part”
should be treated as a separate
> individual. I used to think yes, but lately I’ve been thinking no.
I think that the answer to this question is related to the original
definition of Paul's that started this thread (i.e. reification of
relationships to create identifiable/linkable entities). Back in
October when you and I were butting heads on this topic, I think that
the mistake was to try to demand that an Individual represent both a
physical individual organism and the "reification" (my new favorite
word) of an conceptual entity that connects an Identification to an
Occurrence (a so-called TaxonomicallyHomogeneousUnit) [see Kevin
Richard's comment at
http://lists.tdwg.org/pipermail/tdwg-content/2010-November/001956.html
which first made this distinction clearly]. In some cases, it is easy
to define a resource as being both things (e.g. discrete organisms) but
in others it's not (colonial organisms, clones, herds, tissue cultures,
etc.). After thinking about this for a long time I felt that the best
thing to do was to cleanly separate the physical thing (a.k.a. a
"token") from the conceptual thing (the reified entity onto which we
can hang Identifications and Occurrence records). If one wishes or if
it is convenient, one could describe a particular resource as both the
token and the conceptual thing (e.g. in the case of a discrete
organism) but if it is more convenient one can identify the conceptual
entity as a separate thing and identify any number of related tokens
associated with this entity. These related tokens could span the range
from individual cells to tissues to discrete organisms to herds to
populations and the relationships among them could be described using
dcterms:hasPart or other appropriate object properties. This was the
approach we decided to take with darwin-sw (see
http://code.google.com/p/darwin-sw/wiki/ClassIndividual
http://code.google.com/p/darwin-sw/wiki/TokenIssues and
http://code.google.com/p/darwin-sw/wiki/ClassToken for a more detailed
description of the approach with diagrams and references). This
approach can handle these situations:
Richard Pyle wrote:
> Most people think of a coral head as an “individual”, but it’s
really a colony of polyps. Ants are usually curated as individual
organisms, but can also be useful to collapse to colonies.
Nico Cellinese wrote:
> I also agreed that an individual doesn't have to be whole, so in
my opinion "parts" can also be considered 'the individual in question'.
Similarly, I also agree that up to populations we can still talk about
individuals. However, it is important to be able to link "parts" to
the whole individuals if needed, e.g. is_part_of.
Separate the abstract TaxonomicallyHomogeneousUnit from the physical
tokens and this issue goes away:
Richard Pyle wrote:
> That is my old thinking. My new thinking is that you abstract up
to an
> whole organism when you have a part (e.g., tissue sample), when
representing
> an Occurrence; just as you would abstract down to "Indidividual",
when you
> what to represent Taxon-at-Event for an Occurrence.
In darwin-sw we did not attempt to define a real ontology of tokens
that are physical parts or conglomerations of organisms
(cells/tissues/discrete
organisms/herds/populations/colonies/clones/etc.), but that would be a
good thing to do. It could be done in isolation from the definition of
the "reified" TaxonomicallyHomogeneousUnit.
Steve
--
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