EML mapping question.

[Christopher Jones]

James W Brunt wrote:

> It seems that some information is getting lost with this solution - in 
> the absence of a triple to define the relationship between the 
> citation and the dataset I don't see how the information "this 
> citation is for a paper that cites or contains analysis of these data" 
> is recorded. Enlighten me. 

Yes, I agree, the semantic definition of the relationship is not 
explicit.  RDF does a much better job at defining those relationships 
(potentially as explicitly as the relational model as Peter had 
mentioned.)  Since the <additionalMetadata> tag has a content model of 
ANY, nothing is stopping us from inserting an RDF document in there that 
defines relationships (which Chad pointed out), or any other means of 
semantic mapping/register.  

Nevertheless, in this case, and in many cases of human-consumed 
metadata, there will be some inferring going on, and *how* the 
relationship may be determined will likely be influenced by the 
*presentation* in an application or web interface.

I also wanted to point out that <citation> is strategically embedded in 
a number of places throughout EML, but the semantics beyond a 
parent-child relationship are no more explicit than the 
<additionalMetadata> tag.  

David, thanks for boldly diving in.  I'm encouraged by the strong set of 
schemas that have resulted from contributions from *many* people.  Keep 
implementing and testing...

Chris