[OAI-implementers] RE: [oai-alpha] Re: Fwd: MARC to DC Translator (ALPHA)

Young,Jeff jyoung@oclc.org
Thu, 25 Jan 2001 17:54:37 -0500

(I'm redirecting this from the oai-alpha list to the oai-implementers list
where most of the related discussion has occurred.)

I'm pleased to hear that LC is leading the effort for a one-way crosswalk.
My assumption is that such a specification would be definitive making the
entire discussion of my CORC-based server obsolete. Once the spec is
published, we will want to identify the mechanisms that people can use to
perform the conversions. For example, I can imagine a MARC-XML to DC
transform being performed using XSLT stylesheets. Having such a stylesheet
available would make life very easy for many of us.

Assuming that discussion of my server isn't entirely moot now, you're quite
welcome to run your records through it. I'm confident that you'll find
problems, but that's why I called it an alpha. I promise to resolve any
problems you report as quickly as I can.

CORC's crosswalk is table driven, so, in theory, I should be able to package
it up in readable form. I'll distribute something to the list as soon as I


> -----Original Message-----
> From: Caroline Arms [mailto:caar@loc.gov]
> Sent: Thursday, January 25, 2001 5:19 PM
> To: OAI alpha group
> Subject: [oai-alpha] Re: Fwd: MARC to DC Translator (ALPHA)
> Jeff,
> I'd like to echo Diane Hillman's implication that what matters when
> converting from MARC to Dublin Core is the mapping you 
> choose.  Is yours
> documented somewhere?  I have had some experience with 
> receiving exported
> DC records from CORC and know from the receiving end that it 
> has been a
> moving target over the last six months.
> The mapping we have used for LC's initial Open Archives 
> implementation was
> worked out in conjunction with Rebecca Guenther of the MARC Standards
> Office as a mapping suitable for one-way mapping (as opposed 
> to the CORC
> round-trip mapping).  It also takes into account the choices of how to
> index particular MARC fields and subfields we have made over 
> the years for
> American Memory.  Rebecca has not yet documented her 
> recommended one-way
> mapping publicly (on the MARC Standards web page) but intends to.
> Since I'm not the programmer here (and he's away for a few 
> days), I can't
> run your records through our conversion, but if you are 
> interested, I can
> see how much effort that would be.  Will your system be happy 
> if I run a
> set of 200 MARC records through it?  They will give special character
> conversion a workout.  I'd be happy to send you the results 
> if you wanted
> to do a comparison.
> What we have done for our implementation is similar to Diane's
> mapping, although not identical.  In addition, we decided (at 
> least for
> the time being) to drop a few more fields when converting to 
> unqualified
> DC because we thought systems that only used simple DC would 
> be unlikely
> to cope reasonably with some details -- alternate titles, for 
> example. I
> (personally) have been hoping that practical experience with 
> real data and
> real service providers would guide future practice in the 
> cases that are
> not obvious.
> You can see what our conversion does empirically through Hussein's
> Repository Explorer at the American Memory (lcoa1) 
> "repository."  The set
> of records I would run through your converter would be the Dance
> Instruction manuals.
>      Caroline Arms                                caar@loc.gov
>      National Digital Library Program
>      Library of Congress
> ---
> You are currently subscribed to oai-alpha as: jyoung@oclc.org
> To unsubscribe send a blank email to 
> leave-oai-alpha-156R@www.nsdl.cornell.edu