[eml-dev] [Re: Proposed revision to eml-literature module...]

[Margaret O'Brien]

Sorry to be so late in chiming in. I was on vacation during the second 
half of August, and, since, have been mostly engaged in various local 
emergencies. But one advantage is that I have gotten to read the entire 
thread at once.  Getting the historical perspective has definitely been 
educational, particularly from my POV as an implementer.  BTW, thanks 
for the approval to join this group (I'm particularly excited about the 
salary and benefits that James mentioned).

It seems that generally, Matt, James and Peter agree that creating a 
bibliography schema that imports eml-literature is an acceptable 
solution for maintaining lists of citations, and that a bibliography 
could be considered a resource of a project, individual, or group, so 
EML can be used as it's container.  At first I thought that this was a 
fairly simple addition, but now I see that it carries some implications 
for list-making that could be more precedent-setting.  When I first took 
a stab at moving a bibliography into EML, the choice between handling a) 
hundreds of xml docs plus a results-set or b) a single doc seemed pretty 
clear.  Up until last week, it had been my experience that displaying 
one document from metacat (regardless of length) was much faster than 
waiting for a results-set. However, Chris Jones has installed metacat 
1.6?rc in Santa Barbara, and it's at least10x faster at returning query 
results, so that argument is gone. Eventually, we'll have a programmatic 
method of managing the bibliography, and it wont really matter whether 
there is one document produced or hundreds. Also in the long run, it may 
be more difficult to filter a long xml document based on content, that 
it is to query separate documents and get back a resultset.

So now, I'm starting to think that a more generic <eml-list> that could 
contain any number of any of the basic modules could serve several 
purposes -- you could create a bibliography (either all or a subset), or 
use it to group datasets, protocols and the resulting papers. Possibly, 
using either a harvest-list or resultset accomplishes this already, but 
these both seem to have been designed for other purposes.

This brings up a question I've had about eml-dev's intended use of the 
citation module: Was it intended that the including a citation tree 
within a dataset be used as a mechanism for attaching the appropriate 
datasets to a published paper? Datasets are allowed to contain the 
citation tree in several places, but not at a very high-level.  Some 
journals are beginning to ask that their authors also publish datasets. 
It's possible that the idea of using eml and metacat for this could be 
way off-base -- maybe this isnt the right place for linking a dataset 
and a paper, or maybe it's the responsibility of the publisher, not the 
lter-site.

I've played with one way of attaching datasets to publications (from 
within the citation).  Also, there were some other items in Mark's 
schema that need some feedback from this group. I'd rather bring these 
up in a second email, since this one is already long.  And possibly, 
these are insignigicant in light of the higher level list issue and 
should be put off.

Margaret