[DDI-SRG] Updated 2.1 to 3.0 mapping UPDATE

[Wendy Thomas]

Sanda,

This is the intended mapping based on the definitions in DDI 2.1 and 3.0. 
There are some things in DDI 3.0 that are simply not supported by the 
content of 2.1. Your example of Concept and Universe are a case in point. 
In terms of mechanical translation either the software has to remove the 
duplication during the process automatically or with human intervention OR 
cleanup and modification have to take place after the intial conversion.

If you look at any of the Concept entries you will note that the content 
goes to an entry in the ConceptScheme and the variable or question 
references that entry. If you look at the output of the SPSS and SAS 
coversions to DDI 3.0 you will note that they do not take into 
consideration replication of categories. I've got two options. Use it as 
is and live with a DDI 3.0 complient but not optimized file OR run a post 
translation cleanup. Well a third option is not to use it all.

Clearly translation tools will need to deal with these problems. The point 
of the spreadsheet is to indicate into which object in DDI 3.0 the 
information would go. The biggest problem I was finding was "felxibility" 
of 2.1 in terms of how information was entered in any given field and what 
this means for the translation system.

What I need to know from reviewers is errors in Xpaths, misdirection of 
information to the wrong location and instances of locations where you 
have consistantly interpreted the content of a 2.1 field in a different 
way.

Not sure if this clarifies things for you but keep asking the questions 
and I'll try to answer.

Wendy

On Tue, 13 Jan 2009, Sanda Ionescu wrote:

> Hi, Wendy.
>
> As I look at your 2->3 mapping, I have a "philosophical" question:
> Is this an "ideal world" mapping, meaning - this is how we would map the
> 2.1 fields if we could (or if we did the conversion manually, or
> semi-manually?); or, is it supposed to have a practical application
> (i.e. specs for a conversion tool)?
>
> And this is the reason I'm asking: as I'm sure you know, I attempted a
> mapping myself this past summer, with the practical intention of
> building a conversion stylesheet. So I had to figure out what the
> stylesheet would do. In some instances I found out that even though a
> theoretical mapping was possible, it was not feasible in practice.
> Examples would be "concept" and "universe" at variable level - if we
> created an entry for each concept and universe statement found under
> variables, we would end up with identical entries repeated at nauseam in
> the schemes, with the same content but different IDs! We would also be
> unable to use the in-built hierarchy in the universe scheme, since a
> machine would not be able to establish whether a certain universe
> (expressed in 2.1 as a string) is a child of another, or else. The same
> thing applies to some of the responsibility statements in DDI 2.1 - we
> may have the same organization as author, producer, distributor, and if
> we created organization entries for each time it appears, we would end
> up with a lot of redundant entries in the Organizations Scheme.
>
> So, to me it is important at this point to establish whether we plan to
> release an "ideal" mapping, or a "practical" mapping, because this
> knowledge would inform my evaluation of the current mapping.
>
> Thank you so much.
> Sanda.
>
> Sanda Ionescu
> ICPSR
> University of Michigan
> P.O. Box 1248
> Ann Arbor, MI 48106
>
> Phone, Fax: 734-615-7890
>
> -----Original Message-----
> From: Wendy Thomas [mailto:wlt at pop.umn.edu]
> Sent: Monday, January 12, 2009 5:59 PM
> To: TIC list
> Cc: Sanda Ionescu; tassoukis at iza.org; askitas at iza.org; Mary Vardigan
> Subject: Re: Updated 2.1 to 3.0 mapping UPDATE
>
> In working with the mapping I have already found and corrected the
> following errors:
>
> VersionResponsibiliyt  corrected spelling to Responsibility
> subTitl should be ../r:SubTitle  not r:Title
> altTitl should be ../r:AlternateTitle  not r:Title
> pPhysicalDataStructure should be p:PhysicalDataStructure
> r:FundingInformation/r:Role should be r:FundingInformation at role
> r:GrantNumber/r:Role should be r:GrantNumber at role
> COMPLETED CONTENT FOR 1.2 and 1.3 as follows
> 1.2	guide?	PCDATA
> s:StudyUnit/l:LogicalProduct/l:DataRelationship/r:Description
> r:Description	r:StructuredStringType	Initial home will need to be
> parsed
> 1.3	docStatus?	PCDATA	s:StudyUnit/r:VersionRationale
> r:VersionRational	r:InternationalStringType
>
>
>
> 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
>

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