That was the topic discussed recently by OCLC Research Library Partners metadata managers, initiated by Melanie Wacker of Columbia University, Roxanne Missingham of Australian National University, and Sharon Farnel of University of Alberta. Librarians and administrators are well aware of the tension that exists between delivering access to our library collections in a timely manner and providing good quality description. The metadata descriptions must be full enough to allow us to manage our collections and to support accessibility and discoverability for the end-user. Many libraries need to compromise by using vendor records, by creating minimal or less-than-full level descriptions (according to existing guidelines such as BIBCO or creating their own) for certain types of resources, and by limiting authority work. We need to better understand the impact that these compromises are having on our end users.
Generally, Partners apply full-level cataloging to all metadata created in-house but accept lower-level cataloging from vendors. Minimal-level cataloging is commonly used as an alternative to leaving materials uncataloged, often as a result of large volume of materials and insufficient staff resources. The types of materials receiving minimal-level cataloging include theses, e-resources (generally accepted from vendors as is), ephemera, old backlogs, gift materials, and materials deemed to have low research value. Minimal-level cataloging may be enhanced by adding FAST (Faceted Application of Subject Terminology) headings rather than LC Subject Headings. Minimal-level cataloging is also used as “placeholders” for inventory control, with the hope that they can be upgraded later or when an item is requested. Decisions are often driven by expediency. The metadata created also depends on the format—for example, whether it is for materials in an institutional repository or digital collection. Regardless of the format, everyone strives for consistency and authorized names.
Inconsistencies in vendor records, especially variations in a specific author’s name and different romanizations for materials written in non-Latin scripts, hamper discoverability. Some try to remediate these problems using batch processes such as OpenRefine. The National Library of Australia and Australian National University (ANU) have been working with publishers and publishers’ organizations to improve the consistency in their metadata and to start using ISNIs, as better, more consistent metadata would also benefit their other clients such as booksellers. ANU reported that this advocacy improved the quality of the metadata provided by Knowledge Unlatched.
Metadata specialists consult “collection stewards”—specialists who are familiar with both the collections and users’ needs such as special collection librarians, curators, and archivists—on prioritizing the metadata applied, allocating staff resources, and which data elements are most needed. These priorities may complete with each other. Typically, metadata specialists have no direct contact with end-users. When they receive requests for specific data elements—such as the supervisors for theses—there may not be a way in current systems to handle such role distinctions appropriately.
Many user studies have been conducted, but they generally focus on the discovery layer interfaces rather than the underlying metadata. One exception: The Library of Congress’s Digital Collections Management and Services Division plans to start a project in 2019 that will look at potential effects of different levels of metadata on discovery and use of web archives.
Partners refer to usage data to prioritize clean-up work, weeding, and purchase decisions, but not for reviewing the impact of different levels of metadata. The one exception noted was Cornell’s examination of interlibrary loan records to see if items stored remotely and cataloged according to their minimal level guidelines could be found. They determined that resources that otherwise would have remained uncataloged were being requested.
The current prevalence of keyword searching in discovery layers overshadows the value of controlled access points that collate materials in different languages. Increasing use of identifiers may help bring materials together again in a future linked data environment. It then becomes even more important that the identifiers associated with personal names are the correct ones. Integrating authority record data into discovery environments is another area that needs attention, including reconsidering authority standards’ orientation toward browse functionality.
Migration to a new integrated library system prompts systematic reviews of metadata. It also offers an opportunity to structure more efficient workflows and use the system to review metadata quality so that everyone can better use the catalog.
Karen Smith-Yoshimura, senior program officer, works on topics related to creating and managing metadata with a focus on large research libraries and multilingual requirements.