Message-Id: <199907121957.MAA34594@dns.ccit.arizona.edu> Date: Mon, 12 Jul 1999 15:46:15 -0400 From: Jill Ann Hurst <mailto:jhurst@MCLS.ROCHESTER.LIB.NY.US> Subject: MARC vs. Dublin Core -- summary of responses To: mailto:IMAGELIB@LISTSERV.ARIZONA.EDU
Thanks to everyone who responded to my message.Some notes:
* Although there is some information on CORC at the OCLC web site, it seems to only be available to those who are currently authorized to use it.
* Another respondent said the comparison should be of MARC to XML and AACR2 to Dublin Core. Syntax vs. semantics. Containers vs. content.
I would like to say that the answer is now clear, but I can't. What I can say is that as a non-cataloguing librarian (yes, one of those), I can more clearly see the options and opportunities.
Below are all the responses I received.
Thanks!
Jill (mailto:jhurst@rrlc.rochester.lib.ny.us)
-=-=-=-
At least in some quarters, there is the option to have your cake and eat it. I'm sitting on a local group that's trying to work with the OCLC CORC database, which has a novel characteristic of translating Dublin Core to MARC or MARC to DC at the push of a button. Of course, this drives our authorities person nuts who (correctly, I think) argues that you cannot reliably go from a broader communication standard to a more precise one (DC to MARC). So, in the case of CORC, Minnesota opted to use Dublin Core, believing it to be the lowest common denominator. Since their is really nothing to be gained by MARC cataloging in CORC, why go to the extra bother?
The real issue in such a union database is content standards, which CORC purposely stays away from. We decided to impose our own, minimalist, content standardization (such as inverted names), just so we had a prayer of a chance to retrieve our own institution's records effectively. Just can't let go of some AACR2.
Bruce H. Bruemmer, Coordinator University of Minnesota Libraries mailto:Bytelib@tc.umn.edu http://digital.lib.umn.edu
-=-=-=-
Those interested in guidelines on metadata and scanning standards for collaborative based initiatives please take a look at the Colorado Digitization Project's website http://coloradodigital.coalliance.org. In the ABOUT section you will find information that we've developed on standards.
Jill, now to your question, we've found that we had a variety of approaches already underway in Colorado, some based on MARC, some using DC imbedded in an html record, some using MARC for collection level cataloging and then finding aids imbedded in HTML records, while others using a database like Access. We are taking a distributed networked approach, rather than loading all images and metadata into one system. As we are a cooperative of museums, libraries, archives and historical societies, we decided we couldn't dictate one standards, so what we did was develop a set of core elements that had to be present in all records, regardless of what standard was used.
Additionally we have decided to create a union catalog as a means of improved retrieval of images, as the current approaches (browsers, etc) are satisfactory. We are looking at OCLC's SiteSearch software that can load multiple formats of data. Using OCLC's Mantis software (which will be available as stand alone software and is DC based) we will allow organizations to input records directly, for that we will use DC.
Currently we are participating in the OCLC CORC project.
Hope this helps.
Liz Bishoff The Colorado Digitization Project mailto:BISHOFFL@CONCENTRIC.NET
-=-=-=-
On the topic of having and eating cake, there's also the option of doing MARC records that contain DC's 15 elements. The MARC-DC crosswalk is on the LC site at http://lcweb.loc.gov/marc/dccross.html.
We've been doing rather basic MARC records that contain all the DC info in all the places that DC will expect to find that info (often different than where we would put that same info - our answer is that it appears in the record twice).
paula willey [mailto:pwil@AMNH.ORG]
-=-=-=-
The Library of Virginia Digital Library Program began in May of 1995 before Dublin Core metadata standards were fully evolved. MARC records were created for all of the individual photographic images and linked to the images via the 856 field. We have continued to employ this method for all of the photograph collections we have digitized (approximately 17,000). We also have chosen to use MARC to create records for individual digitized text documents (approximately 103,000). The reasons we chose MARC are as follows:
1) Portability of the MARC format across library automated systems. (The LVA DLP images are available through a web Gateway which is part of our library automation software.)
2) Flexibility of MARC record. We have used full-standard MARC records to describe graphical images and item-level archival documents. We have also used MARC records which we have "adapted" in order to describe other entities such as index entries.
3) Stability. MARC is a known, stable standard. It is not likely to be abandoned in the foreseeable future.
4) Interoperability. Other emerging standards, both metadata and searching/indexing, will likely take into account the vast number of MARC records currently in existence. Crosswalks are an example.
We also use Dublin Core metadata to describe the web pages we create for each collection we have digitized. This includes a number of collections where MARC records were not created such as for the Electronic Card Indexes. We currently have about 70 different collections, indexes, or finding aids that are considered part of the Digital Library Program. The metadata elements were added to the collection web pages in order to facilitate retrieval by any Internet search engines that recognize these elements. The search engine on the Library of Virginia web site is not very robust so that was not a factor in adding DC metadata. Also, there is not at present any searching capabilities across all the DLP collections. That is something we will hopefully be implementing in the future as some of the new standards (e.g. XML) become stable.
We have not cataloged the sites for the various digitized collections, primarily due to local factors. However, if these sites were cataloged, the records would be available in our OPAC, in OCLC, in InterCAT, in FirstSearch, and possibly in NetFirst which would increase access.
The URL for the LVA DLP is: http://www.lva.lib.va.us/dlp/index.htm
This, at least, has been our experience.
Jean Marie Taylor [mailto:jmtaylor@vsla.edu