[DDI-users] DDI usage scenario

[Wendy Thomas]

Alerk,

You've raised some interesting questions, some of which need further 
discussion in TIC. The five questions that immediately come to mind or 
that were raised in an offline discussion are:

1) What is the appropriate use of Name and Label outside of the Variable 
contenxt? This is important because systems will either be looking in the 
Label field or Name field and their use needs to be consistant and 
interoperable.
2) The use of Variable Group for grouping variables by concept. In general 
this is not a practice we want to encourage. Variables link to concepts 
directly and this link should be used to build these "groups". However it 
is appropirate as a means of relating a group of variables to a subject 
heading or keyword.
3) The definitiion of Concept. DDI 3 is much tighter on its definition of 
concept and we need to reinforce the differentiation between concept, 
subject, and keyword. A Best Practice is currently under development on 
this.
4) Why is Universe required? It doesn't seem to make sense in this 
context. RESPONSE: Universe is required in any study. A study needs to be 
about something. The only way around this is to publish not a study but a 
ResourcePackage of ConceptSchemes and VariableSchemes (containing 
VariableGroups).
5) Do the variable reference presume that there is a DDI 3.0 file with the 
full variable information somewhere else? What if that is not the case? 
Can the text of the variable and related information (question, category 
scheme etc.) be put in this study? RESPONSE: DDI references explicitly 
state "Used for referencing an identified entity epxressed in DDI XML..." 
so it needs to be in DDI 3.0. If not available in DDI you could enter the 
content of the variable and then reference it. (VariableGroups only 
include by reference).

A basic consideration here is whether this is really a study. I don't 
think we want to loosen up the requirements for a Study to support 
alternative applications. All of the parts used by the system Alerk 
describes could be published as a ReferencePackage containing a 
ConceptScheme and VariableScheme.

Its an interesting case study because it pushes DDI further along the 
lines of supporting new applications and new approaches to information and 
information capture. The items noted above are being placed on the TIC 
agenda for discussion in one of our next meetings. I'll make sure and 
relay a summary of that discussion back to this list.

Wendy


On Fri, 6 Feb 2009, Alerk Amin wrote:

> Hello,
>
> We are are currently trying to use DDI 3 as the output for a
> documentation tool that we are working on. Our needs do not fall into
> the normal "lifecycle" model, and we are running into some issues. We
> would appreciate any help or insight about how we can use DDI with our
> project.
>
> First, a short description of our project. We are building a website that 
> allows researchers to document their publications and the concepts that they 
> use in the publication. A example of a concept in a article would be 
> 'household income' referenced by a variable in the dataset that contains the 
> actual household data values. Another example would be 'active social 
> participation', measured through a group of variables.
>
> In our project we use a datasets that reside in data archives and for which 
> pre DDI 3.0 documentation is available through Nesstar. Researchers typically 
> only use a selection of variables from a dataset in their publications. We 
> want to capture and reference only the variables that the researcher used and 
> the concepts that - as the example shows - can be tied to one or more 
> variables.
>
> Using our website, researchers can create concepts that they are studying in 
> their publication, and then select the variables they used to measure that 
> concept. We want to create a DDI instance for each publication that captures 
> the additional metadata for that publication. In this case, that would be the 
> list of concepts and the list of variables for each concept (in case a 
> concept is represented by more than one variable). The concepts are "new" 
> information, but the variables should link back to the original XML for the 
> dataset.
>
> The attached sample.xml file contains a small example showing how we think 
> this information should be captured with DDI for one publication.
> Does this seem like an appropriate solution? Or should we use a different 
> method for placing this information inside DDI XML?  We would appreciate any 
> comments or suggestions. Thank you for your help.
>
> Best,
> Alerk, Rob
>
> -- 
> Alerk Amin
> Senior Software Developer
> CentERdata
> Room K737 (Tilburg University, Koopmans Building)
> Postal address : PO Box 90153, 5000 LE Tilburg, The Netherlands
> Visiting address : Warandelaan 2, 5037 AB Tilburg, The Netherlands
> Telephone : +31-13-466 2243 / 8325
> Fax : +31-13-466 2764
> WWW : www.centerdata.nl
> Disclaimer : See http://www.centerdata.nl/maildisclaimer.
>

Wendy L. Thomas                          Phone: +1 612.624.4389
Data Access Core Director		 Fax:   +1 612.626.8375
Minnesota Population Center              Email: wlt at pop.umn.edu
University of Minnesota
50 Willey Hall
225 19th Avenue South
Minneapolis, MN 55455