Review Comment:
The paper describes extremely important work that is of great value to
the biodiversity community. Providing an RDF version of DwC and a
guide on how to use it is direly needed. The guide contains valuable
information and is a very solid piece of work.
Unfortunately, this is not really reflected in the paper submitted to
the special issue. I am not quite sure about its value to the
community in the current form. I am sure, however, that the
experiences and lessons learned from the work on the RDF guide offer
plenty of material for a great paper - and I would really love to see
that paper in the special issue.
Ideally, such a paper would be of interest on the one hand to people
interested in using the RDF DwC version and on the other hand to
people interested in "translating" non-semantic vocabularies to RDF or
OWL.
This would be the case, if the paper explained on the one hand as it
already does to a certain degree the rationale behind some of the
decisions made in the RDF guide, and on the other hand provided some
"lessons learned"-like discussions on pitfalls and difficulties.
Ideally, this should be generalisable to other endeavors.
Some more detailed comments that might or might not still be relevant
after restructuring the paper:
The title is somewhat misleading. One might think, that this actually
is the RDF guide. Maybe rather "Experiences in creating..."
"Lessons-learned from..." "An introduction to" ...
The same impression can be gained from the abstract, so that should be
somehow adapted, too.
Introduction:
"over 428 million Darwin Core records": For SWJ readers not familiar
with DwC: What is a record? Maybe explain using database terms.
"a single table of rows and columns": Are there other kinds of tables?
Section 2:
Figure 1:
Are the objects of the dwcuri:recordedBy triples necessarily refering
to the same realworld entities as the entries in the dwc:recordedBy or
could both be mixed? To use the example: Am I sure that there where
two people recording this or could it have been four? The same
question applies to the other constructs where dwc and dwcuri are used
in parallel.
2.2.1 You state ".. ID terms should not be used as predicates in RDF
triples". And how should the triples look like?
Section 3:
You write in Sec 3.1 "are too complex to suggest ..": Why? what makes
them complex?
Fig 7: If possible highlight the parts where the serialisation differ.
Figure 4 is badly placed.
|