REVIEW OF DESCRIPTIVE METADATA SOLUTIONS

 INTRODUCTION

 Researching other institutions’ approaches to the issues surrounding local descriptive metadata for visual/digital resources should help inform Yale’s own investigation of these problems. What follows under ‘Site Summaries" are some of the more interesting findings. Complete documents may be found at the URL’s listed. These summaries are excerpts from the cited full documents.

  CONCLUSION

Not surprisingly, the most striking conclusion from this review of other institutions’ handling of descriptive metadata issues is that we are not alone – everyone is confronting the same issues of content consistency, i.e., headings; naming, i.e., persistent locations; a lack of fully established record standards; and interoperability between systems and record schema. The ALCTS Taskforce report recognizes these issues.

Solutions to these problems vary. Most sites are still grappling with the issues of uniform headings and persistent locations. They are attempting to resolve issues of record standards and interoperability, but their solutions vary.

• Cornell’s group recommends using MARC records and their local NOTIS catalog.
• Columbia is building a separate relational database using a local record definition that maps to MARC, Dublin Core, etc.
• Harvard’s study acknowledges the limitations of the Marc record as descriptive metadata for visual resources, recommends establishing a core set of fields and a specific thesauri, and advocates the creation of a union visual resources catalog.
• Making Of America II project is using Marc for their descriptive metadata.
• Taking a different approach the Monticello project is using the Dublin Core record as a bridge and Z39.50 as a protocol between separate databases of MARC, GILS an EAD records.

  SITE SUMMARIES

ALCTS Taskforce on Meta Access, Final Report, April 3, 1997, http://www.lib.virginia.edu/alcts/about/final.html

• This taskforce was charged to: … Define Bibliographic Access in the Electronic Environment "to lead the way in defining access and bibliographic control mechanisms for information in electronic form and communicating that mechanism to the users of the electronic information." ... The Task Force has, first, identified problems that are important to the library community and, second, proposed actions by which the library community can influence, participate in and respond to solving the problems.

Of interest to Yale’s Metadata Group are four of the seven identified problems (the remaining problems are education, outreach and environmental scanning. Several of the Actions associated with the metadata problem are also informative. These sections, taken directly from the ALCTS report, follow:  

1. Naming: The need to have permanent and unique names for digital resources is critical. Naming schemes (URNs) should be adopted and used to create transportable, persistent locator information. Two schemes for Uniform Resources Names (URNs) are in use (PURLS and Handles) and others are proposed.
2. Metadata: There will be multiple systems of metadata for purposes of description, discovery and retrieval. Our role is to promote a coherent view of metadata from a library perspective.
   • Action 2C) That current work to develop standards for a simple resource description record for Internet resources be supported and encouraged by the library community, including ALA committees such as CC:DA and MARBI, and that we support the Dublin Core as a starting point for resource description in information production and description activities.
   • Action 2D) That the library community promote the use of metadata standards, especially by libraries, universities and government bodies, in their production of digital documents, but also by other information producers.
   • Action 2E) That ALCTS and the library community promote the use of existing classificatory frameworks (DDC, LC, NLM) to indexing and cataloging agencies.
3. Metadata Syntax and Formatting: Metadata comes in many forms from many sources. Records containing this metadata are often combined in a single local system or moved across platforms. Local systems in libraries need greater flexibility in library catalogs to support continuing use of MARC formats and to accommodate multi-dimensional and hierarchical relationships between/among information resources. The Task Force commends and encourages the ongoing development of MARC/SGML mapping by the Library of Congress and others which will allow consistency of use and the testing of new approaches to bibliographic meta description by libraries.
4. Metadata Records Management: At the Warwick Metadata Meeting (April 1996), there was much discussion on the question of how metadata and objects will be managed locally by library online systems. Local systems include data records from many sources in a variety of formats/syntaxes beyond MARC and the system therefore needs to manage coherently these records for storage, retrieval, display and ongoing maintenance.

COLUMBIA, (http://www1.columbia.edu/sec/cu/libraries/inside/projects/metadata/)

Columbia is implementing a relational database called the Master Metadata File Database to which they will load all their external (e.g., USMARC, VRACore, keyed) descriptive metadata. records. This single metadata database is separate from the library’s NOTIS OPAC. It currently generates HTML, but in future phases it will create USMARC, Dublin Core, and SGML products.

The Columbia Master Metadata File record structure is well defined and is mapped to other metadata schema including USMARC, Dublin Core, SGML and the MESL data dictionary. See: http://www.columbia.edu/cu/libraries/inside/projects/metadata/registry/

CORNELL
Metadata Working Group Report to Senior Management, July 1996, http://www.library.cornell.edu/DLWG/MWGReporA.htm

Cornell’s Metadata Working Group was charged in Oct 1995 as follows:

• Determine the minimum information necessary to include in metadata records to support effective access to electronic resources. These requirements may vary according to the genre of resources. 2) Determine whether this information can be included and effectively accessed in a NOTIS OPAC record. 3) Recommend whether to: a) create a separate database of metadata records for electronic resources not linked to the OPAC; b) create metadata records in the OPAC for electronic resources; c) create a separate database of metadata records for electronic resources that have been copied from the OPAC so that records reside in both places.

Similar to Yale’s short-term approach, the Cornell group, while recognizing the importance of structural metadata, narrowed their study to descriptive metadata. Their study methodology was also similar to ours - Cornell: a) reviewed selected established and emerging standards, and b) inventoried existing Cornell University Library repositories.

From their inventory, the group found, " there is no consistent policy for selecting and cataloging electronic resources within CUL." Furthermore, they found that although, "a great deal of intellectual effort has been expended with the goal of providing access to electronic resources. There has been, however, no comprehensive plan to achieve this goal and unfortunately, the result is duplicative effort, incomplete listings, and occasionally, misleading information."

The CUL library metagroup made three recommendations. To quote from their report, they are:

1. That CUL utilize the data elements which are listed in Appendix B, Minimal Data Elements (see below).
2. That the metadata elements needed for effective access to electronic resources be contained in the envelope of the MARC record.
3. That records be created and maintained in a single database rather than in a separate database of electronic resources. They further recommended that the single database should be their NOTIS system. They wrote, "in the same way that the ISSN provides a critical link between citation databases and the online catalog, the NOTIS ID can provide a linking field between the online catalog and auxiliary databases such as those used for the full text projects."

Cornell’s list of Minimal Data Elements:

Identifier	Numeric or other data that uniquely identifies the object. Date record was created
Title	Subtitle Alternate titles
Author[s]	Photographer Artist Compiler Other names associated with the work Sponsors Translators Illustrators Agents responsible for format conversion, or in other technical roles
Version	Edition Date/Revision date Details that distinguish this version from others Coverage: what library provides with this file How often the file is updated
Source (if reproduction or translation)	Bibliographic information related to the source document
Publication/Availability	Publisher (including address, if applicable) Restrictions on access, sale or use Details for acquiring/accessing the item
Physical description/Extent	Physical format Size

HARVARD
Visual Resources at Harvard University Task Force, http://www.peabody.harvard.edu/VRTG/

Harvard’s task group on Visual Resources was formed in Feb 1997. The operated under the following charge. See: http://www.peabody.harvard.edu/VRTG/appendicesa-b.html

• Harvard's archives, libraries, and museums are actively involved in a variety of initiatives to develop inventory control systems for collections of visual resources, including photographs, posters, prints, paintings, sculpture, and other graphic images and 3-dimensional objects; and to provide access to these collections. In some cases, in addition to descriptive data, digital images of original materials have been created. The provision of access to visual resources, whatever their nature, presents complex challenges including the need to establish standards for description, to determine how records will be structured and displayed, to achieve adequate image capture, and to develop capability for storing and delivering digital surrogates. … In order to answer the question, "What activities are best undertaken collectively to reach desired goals?," the Task Group on Visual Resources has been established by the University Library Council.

  The Task Group shall identify those activities regarding which there is general agreement that progress is best made, and/or optimal results achieved, through a collective effort. An assessment of the timeliness of undertaking specific cooperative initiatives shall also be made, and possible approaches to follow-through recommended. Among the areas to be explored by the Task Group are:

• The nature of the catalogs currently being created at Harvard and Radcliffe for collections of visual resources: whether there appear to be advantages to achieving a certain degree of commonality, and whether there is demand for an off-the-shelf, supported cataloging system.
• The need for providing campus-wide access to visual materials.
• Digital conversion of visual resources to electronic form: accomplishments, interest in continued activity, and challenges to be met.
• Image storage, on-line access, maintenance, backup, and other repository functions.

Similar to Yale, the Harvard group found:

• Some percentage of their visual resources collections were currently automating
• Most collections were creating discipline specific databases and only a few are sharing record structures.
• None of the current projects are Z39.50 compatible.
• They also note that MARC is not widely used in the museum and other visual resources communities both because of a lack of familiarity with the standard and because it does not adequately handle multiple versions of an item nor does it handle administrative (e.g., rights management) issues.

Their conclusions are summarized here http://www.peabody.harvard.edu/VRTG/statement.html

• … the committee issued a report that described a plan for achieving unified access to the University's visual resources by developing a Visual Resources Union Catalog. The establishment of a union catalog will foster and support cross-disciplinary access to visual materials in libraries, archives, and museums, regardless of the physical formats or locations of the originals. The catalog will finally make visual resources at Harvard and Radcliffe accessible as a cohesive resource in the same way that HOLLIS provides unified access to textual resources. The catalog will include descriptive data and links to digitized images.

  The union catalog will be based on the following precepts:

  • the establishment of a core set of fields to accommodate collection-level and item-level descriptions;
  • agreement on descriptive standards, including the use of specific thesauri; and
  • the development of standard methodologies for describing visual materials.
  • The catalog will be centrally maintained. In some cases, the records will be derived from descriptive information in local collections management or public access databases. In other cases, the catalog will serve as the primary source of descriptive information about visual collections for repositories with no local collection management systems of their own.

MONTICELLO ELECTRONIC LIBRARY PROJECT
http://www.solinet.net/monticello/monticello2final.html

The Monticello Project’s purpose was to link distributed regional resources. Participants are geographically located in southeastern US and include public libraries, academic libraries (special collections) and state agencies. Records from these participants are coded in: MARC, GILS and EAD. OCLC provided extensive technical support for this project.

The Monticello Project solution to access to disparate datasets is "to demonstrate that commonly available technologies can be used to network distributed information resources in widely varying environments using standards was accomplished. Using the Dublin Core metadata elements as a bridge and Z39.50 as a protocol, users are able to search across disparate MARC, GILS, and EAD datasets."

Lessons learned from implementing this system include:

• Linking distributed information resources using commonly available technology and standards is a feasible undertaking for not-for-profits ... There is enough development of software, standards and organizational capability to enable linking of distributed information.
• The technical issues of linking disparate and distributed information resources are complex, but not by any means insurmountable ... Technical issues became a problem only when there was insufficient access to educational resources or when technical staff resources are scarce.
• Building in the capacity to access information in several different data formats adds complexity, but is possible given strict definition and adherence to standards, and given that participants are willing to work cooperatively.
• The usefulness of a common user interface will be undermined if any one of the distributed sources of information fails to adhere to the definitions, standards, and cooperative agreements. If a common interface to several distributed databases is desired, one unanticipated outcome is that the display of data is sometimes forced to accommodate the least robust datasets. What initially seems like allowing flexibility in data structure for participants may result in a lack of flexibility in system functions.

Information regarding the Monticello navigation model is available from the Annual Review of OCLC Research, http://www.oclc.org/oclc/research/publications/review96/mont.htm

MOA II
The Making of America II Testbed Project White Paper, Version 2.0 (September 15, 1995), http://sunsite.Berkeley.EDU/moa2/wp-v2.html

Although useful for discussions about Structural and Administrative metadata, the MOA II project intentionally did not focus on descriptive metadata.

For descriptive metadata, the MoA II project is building a union catalog, with MARC records contributed by the participants. EAD encoded finding aids are also being used. Researches will search the union catalog of collection level MARC records that are linked to finding aids.