To  : Meta2 listserv participants
From: Karen Hsu, The Research Libraries of The New York Public Library (NYPL)


      It was in late July that I joined the Meta2 listserv and have been 
following the discussions here, or trying as best as I can to follow. Please 
forgive me if some of the questions raised below have already been addressed.

      I understand that the purpose of the Dublin Core elements is to describe 
electronic resources. There are two kinds of electronic resources, those 
created on the Web and not derived from sources in other formats, and those 
derived from sources in non-electronic formats. It is within the second 
category that I have questions, more specifically, questions regarding 
digitized materials.

1. "SOURCE" element

     NYPL has digitized a few collections and is getting into more digital 
projects. We, as everyone else, have been grappling with metadata issues and are
experimenting with DC elements for some of the projects. An issue arose early 
on, i.e., where to enter description of the source material. The basic principle
of DC is to describe the item in its present form, meaning the electronic form. 
There is an element, SOURCE, reserved for the description of the source. And 
here the problem occurs. It seems the description of the source is all lumped 
together here. For instance, the date of publication (or production) of the 
source, the place of publication of the source, and the publisher of the source 
are all entered here in free text form. Many of these source materials do not 
have online cataloging records, let alone ISBN or ISSN, for linking.

     One of our digitized collections includes photographs and illustrations 
produced in the 19th century. The date is of crucial importance. However, 
according to DC definitions, "1997" should be the value entered in the DATE 
element, because 1997 was the date the resource was made available in its 
present form. It would be quite misleading to have the present date listed as 
the date of a 19th century photograph. Granted, we could add a display constant 
(or qualifier) such as "Date.digitized" to the 1997 DATE. But the date of 
publication of the source would be entered into SOURCE element and would not be 
indexed as DATE, and therefore, not retrievable by DATE when date of the source 
would also be an important access point. The same problem goes for publisher of 
the source and format of the source.

2. Multiple Version problem with MARC format 

     The Problem described above is similar to the multiple version issue 
with the MARC format. I am somewhat sorry to raise the issue now, but DC will 
have to face it sooner or later. The most well known problem with multiple 
version issue is how to describe a microfilmed item. AACR2 (Anglo-American 
Cataloging Rules, 2nd ed.) prescribes that the microfilmed item be described 
in the body of the cataloging record and that the physical description of the 
original be entered into one single note. Serious service problems spring up 
when a cataloging record for a 19th century item has a late 20th century 
microfilmed date, place and publisher in prominent areas and crucial 
information about the original material is all buried in a note.

     To address that problem, the Library of Congress made a Rule 
Interpretation (LCRI) and changed the rule around. It states that, for a 
microfilmed item , the original material is to be described in the body of a 
cataloging record and the description of microfilm features is put in a note 
(533 field). However, in order to indicate to library users in a prominent 
area that the item is on microfilm, LCRI also adds the term "microform" as a 
General Material Designation in the Title Field (245). And technical 
information about microfilm production is entered into MARC's Fixed Fields. 
The result is a description of two formats in one record, a rather unwieldy 
practice and a dubious cataloging principle. But it seemed to have served the 
practical needs of many libraries.

3. A Suggestion

     In a way, the DC elements follow the AACR2 principle rather than the 
LCRI. And that is the problem NYPL has encountered in its experimenting with 
DC. I am not advocating that DC follow the LCRI practice. Rather, perhaps this 
is the time to explore a new structure. The problem of multiple version, in my 
opinion, is more with MARC format than with AACR2. MARC has served the 
nation's (and the world's) cataloging needs remarkably well for the last 
thirty years, but it has its limitations. It is linear and flat, and cannot 
handle different levels of data very well, if at all. Since DC is a new 
paradigm and is still being developed, we should seize the opportunity to 
address MARC's shortcomings.

     My proposal is for DC to make the SOURCE element a level by itself, with 
other elements available for its appropriation. In other words, under the SOURCE
element, one can repeat the DATE, PUBLISHER, FORMAT, or any other elements 
needed to describe the source material. This would be a deeper and different 
level of description and yet attached to the present level. This way, not only 
all the data about the source could be indexed, it could also be easily 
displayed together. 

     To illustrate my point, I am borrowing Ricky Erway's examples from her 
7/18 message. Her examples provided one brief DC record for a Web page 
created by David Phillips on Michelangelo's David and one DC record for the 
sculpture by Michelangelo. According to my proposal, the record for the 
sculpture would become SOURCE of the Web page record and the two records 
would be combined into one as follows:

Title:           Michelangelo's David [that's the name of Phillip's page] 
Resource Type:   web page
Creator:         David Phillips
Date:            1996-04-15
Format:          html
Publisher:       David Phillips [or Onramp Access, Inc]

<Source:
Title:           David di Michelangelo [or whatever M named it] 
Resource Type:   sculpture [or in more general terms, object?] 
Creator:         Michelangelo 
Date:            15xx [or whatever]
Format:          stone-carving [or whatever] 
Publisher:       Galleria dell'Accademia>


     I understand that several issues would be involved here, systems design and
browse display among them. And the most sticking question could be the 
interoperability, e.g., mapping into MARC. Since MARC is one level description, 
it would be difficult to map DC into MARC if DC incorporated different levels of
description in one record. But then, if DC is to be kept completely congruent 
with MARC, it would probably not be able to resolve the limitations of MARC. 

***************************************************************************** 
The opinions expressed here are personal and do not necessarily represent the 
official views of The New York Public Library

Karen M. Hsu
Assistant Director for Cataloging
New York Public Library
42nd St. & 5th Ave.
New York, N.Y. 10018-2788
(212) 930-0702
(212) 930-0785 (Fax)
Internet: [log in to unmask]
******************************************************************************