linking protocol to dataset

[Peter McCartney]

With respect to the first question, im going to turn your own distiction of
prescription and description back on you. We agreed that a protocol was a
prescription - an abstract methodological procedure that is independent of
any particular data set. its a statement of how to do something. a dataset
exists and thus must have had something actually done at some point. that is
your method - a statement of how did something. so the way you bind a
protocol to a dataset is to say "i did this (method) following that
procedure (protocol) - pretty much as you say. Ifyou used different
protocols at different points in the process, then you put in multiple
method steps. the only potential problem i see is if we consider that the
repeating methodSteps must be sequential steps within a single operation
rather than a collection of methodological operations that might be
relatively independent of each other as describe. But it seems to me that
with the ability to nest substeps, plus the section element in text type,
theres plento of opportunity to accomodate a resonable range of nesting.

with respect to the confusion over repeating the choice versus repeating the
steps, i think this is a schema design error. I think the repeating choice
should have been a non repeating sequence. its an error to make a choice of
optional elements. 

Peter McCartney (peter.mccartney at asu.edu)
Center for Environmental Studies
Arizona State University
480-965-6791 

> -----Original Message-----
> From: Tim Bergsma [mailto:tbergsma at kbs.msu.edu]
> Sent: Wednesday, October 09, 2002 9:08 AM
> To: Peter McCartney; Eml-Dev (E-mail)
> Subject: linking protocol to dataset
> 
> 
> Peter, others,
> 
> I'm wrestling with the relationship between a protocol and a dataset.
> 
> In my local data management system, a dataset can 'have' an associated
> protocol.  But in eml, a dataset cannot have a protocol:  it can only
> have a methods that has a methodStep that has a protocol.  
> 
> In fact, in my local system, a dataset can have any number of named
> methods, and any number of named protocols.  But eml clearly expects
> that a dataset has only one methods.  The only way I can see to
> stipulate multiple, independent methods is to create multiple
> methodSteps, and poke the descriptions into TextType.  Same for
> protocols.  Clearly this was not how methodStep was intended.
> 
> MethodStep repeats, but also the choice of
> methodStep/sampling/qualityControl repeats.  Correct me if I'm wrong: 
> choosing methodStep twice will look exactly like choosing it once but
> having two methodSteps.  Consequentially, there's no 
> mechanism to signal
> that one sequence of methodSteps is independent of a second 
> sequence of
> methodSteps.
> 
> Unless review of RC2 flags this as a certifiable flaw, it's 
> not going to
> change now.  I have a very real problem that some of my datasets don't
> have records of actual activity (methods), they only have 
> protocols.  I
> could spoof the methods/methodStep tags and poke each protocol into a
> textType.  For datasets with multiple, independent methods, I 
> guess I'll
> have to poke each into a sequential methodSteps, even if they are not
> sequentially related.  Please comment.
> 
> Tim.
> 
> 
> 
> Tim Bergsma
> LTER Information Manager
> W.K. Kellogg Biological Station
> Michigan State University
> Hickory Corners, MI   49060
> 616/671-2337
> tbergsma at kbs.msu.edu
> http://lter.kbs.msu.edu
> 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mercury.nceas.ucsb.edu/ecoinformatics/pipermail/eml-dev/attachments/20021009/c674268f/attachment.htm