E-book collections are growing, and many libraries are integrating e-book records into their online catalogs for ease of access. Given the large size of many e-book packages, libraries often use vendor-supplied record sets to expedite access. A review of the literature reveals that while numerous publications have addressed issues of access and bibliographic control of e-journals, little research has been done in the area of e-book cataloging, particularly for mass cataloging and management of vendor-provided e-book records. Only a few publications discuss cataloging rules, case studies, or survey results pertaining to e-books.

In a 2006 paper, Sanchez and colleagues shared techniques for batch editing and maintenance processing e-book records using nontraditional editing utilities.⁵ The authors identified problems in bibliographic records provided by NetLibrary and described efficient record-editing methods to clean up NetLibrary cataloging records. The paper documented procedures for error resolution using a variety of tools, including MarcEdit, Microsoft Word macros, and Microsoft Excel spreadsheets. Sanchez and colleagues pointed out the need to establish workflows and “create procedures that detail a step-by-step approach to editing and revision tasks.”⁶

Bothmann discussed e-books as unique manifestations in his guidelines for e-book cataloging.⁷ His article described in detail the functional elements of cataloging e-books using the 2002 revision of the Anglo-American Cataloguing Rules (AACR2). He pointed out those areas and fields to which catalogers should pay special attention when cataloging an e-book, including control fields, variable data fields, uniform titles, title information, edition information, type and extent of the resource, publication and distribution information, physical description, series statement, notes, and subject analysis. Bothmann recommended that catalogers make good use of cataloging rules and keep up-to-date with current rules.

Martin addressed e-book cataloging questions, such as where e-book records should come from, how to process them, how to handle holdings, what changes to make to vendor-provided records, how to maintain e-book records, and whether to add holdings to OCLC.⁸ She raised concerns particularly about the limitations of applying the electronic-reproduction model when using vendor-provided records in e-book cataloging. According to Martin, e-book cataloging “is not a simple task and requires careful analysis and thoughtful decisions.”⁹

Simpson, Lundgren, and Barr from the University of Florida Smathers Libraries (UFL) described efforts to enhance access to print and electronic versions of the same title within the catalog by linking corresponding manifestations.¹⁰ Using a local loader, Excel spreadsheets, and macros, their Functional Requirements for Bibliographic Records (FRBR) conceptual project employed a highly automated, multistep process to identify, match, and link netLibrary e-book records with their print counterpart records. This linking model helped users to effectively search and retrieve both versions of the same title by taking advantage of keyword searching of the table of contents data acquired by UFL in 1990 to enrich records for print books. According to Simpson, Lundgren, and Barr, this FRBRizing model can be applied to link records of other related materials. The authors recommended that catalogers “go beyond their traditional functions, explore new options in technology, and communicate their ideas to those who can implement them and to those who benefit from the outcome.”¹¹

Belanger’s 2006 survey examined the cataloging practices of thirty higher education libraries in the United Kingdom.¹² The analysis of the survey results shows that most of the libraries cataloged e-books from large subscription collections. Five libraries cataloged individual e-books while only four libraries cataloged free e-books. Only two libraries had not cataloged e-books at all. Twenty-three of the thirty libraries used separate records for print and electronic versions of the same title. The survey also indicated that very few universities’ online public access catalogs (OPACs) allowed retrieval of e-books by limiting the search to the e-books format. Belanger concluded that “much work remains to be done in order to ensure easy access to electronic books via the library OPAC.”¹³

Writing in 2000, Hawkins noted that “the ebook market is in a state of extreme flux and is changing daily.”¹⁴ The same is still true a decade later. At every level of the e-book landscape, “fragmentation, of technical platform, of format, of business model … complicate service provision.”¹⁵ E-book providers regularly emerge and disappear as publishers and content aggregators change hands. As the e-resource marketplace has matured, a wide variety of monographic e-content has become available, including reference works, academic and technical books, and literary and primary source content. The means of obtaining access to e-books are similarly diverse, including single-title purchasing through traditional fulfillment services, publisher and aggregator packages, and user-driven acquisition. This complexity is a challenge for catalogers not because the resources are difficult to catalog, but because the workflow is difficult to manage efficiently. E-book collections are volatile, and the bibliographic data that support them come from many places and follow few standards.

In August 2009, the PCC Provider-Neutral E-Monograph Record Task Group issued its cataloging guidelines for e-monographs.¹⁷ Like the aggregator-neutral policy adopted for electronic serial records in 2003, the PCC e-monograph record policy adopts the model of a single master record encompassing all equivalent manifestations of an e-monograph title rather than separate records for each provider’s version.

The provider-neutral approach has two significant benefits. In a shared cataloging environment, like the WorldCat database, the provider-neutral approach halts the proliferation of incrementally different records for the same content. The extent of this problem is best illustrated with an example. According to the SerialsSolutions knowledgebase, Richard L. Shell and Ernest L. Hall’s Handbook of Industrial Automation (Marcel Dekker, 2000) is available online from five different providers and is a component of more than a dozen packages. A search for this title in WorldCat yields twelve records for online manifestations, mostly duplicate records for the same two versions, one from ebrary and another from ENGnetBASE, both of which are part of UHL’s e-book collection. Each provider has exposed slightly different bibliographic metadata, but the Handbook is the basis of all of them. In UHL’s experience with aggregator-neutral serial records, fewer and more consistent e-resource records in the shared database have made finding and identifying appropriate records much easier for the cataloger. Effort once spent sifting through numerous similar records for the best match or inputting new records that closely replicate existing ones can be devoted instead to enriching the master record with subject headings, contents, and authority work. The new policy also does away with the distinction between reproduction and born-digital monographs and provides clear instructions for the use of fields that were previously applied inconsistently, such as 534 (Original version), 773 (Host item entry), and 776 (Other format).

Under the new guidelines, dates and publisher information are based on the original monograph, whether print or electronic. This change harmonizes cataloging practice with what UHL has discovered to be a user preference for seeing the original publisher and date information in the publication area. Most note fields have been eliminated, with the exception of a suite of notes that pertain to electronic manifestations emanating from the Digital Library Federation Registry of Digital Masters and other digital preservation projects. Archival digital masters are the only allowable use of the 506 (Restrictions on access), 533 (Electronic reproduction), 538 (System requirements), and 583 (Action) fields in master records. Figures 1–3 illustrate the difference between previous cataloging practice and the provider-neutral approach, again using the Handbook of Industrial Automation example. Figure 1 is for a version cataloged as a reproduction and figure 2 is for a hypothetical version cataloged as a born-digital manifestation. Figure 3 shows the same title cataloged according to the new provider-neutral guidelines. The major areas of difference include the following fields: 020 (ISBN), 260 (Publication, distribution, etc.), 300 (Physical description), 533 (Electronic reproduction), 710 and 730 (Added entry), 776 (Other format), and 856 (Electronic location and access). The basic description in figure 3 applies to all known online manifestations. Version-specific notes and access points, such as system requirements, reproduction information, provider and package name, and URL notes pertaining to library-specific access that appear in figures 1 and 2, are now considered local data and are not to be used in the provider-neutral master record. The record in figure 3 could replace all twelve records for electronic manifestations of this title that currently exist in WorldCat. The Provider-Neutral E-Monograph MARC Record Guide provides detailed coverage of the new encoding rules.¹⁸

While the provider-neutral e-monograph policy will bring greater structure and coherence to e-book cataloging, challenges remain for local batch record management. Few vendors have converted their existing records to a provider-neutral state. Hundreds of thousands of vendor records continue to be issued with reproduction notes and package and provider names, or are cataloged incorrectly as born-digital editions. Local implementation of the provider-neutral guidelines is not required even for PCC member libraries, so individual libraries must decide whether to convert existing records to the new standard now, wait for their data providers to make the change, or ignore the changes altogether. UHL is pursuing a gradual implementation that will bring provider-neutral records into the catalog at the time of their regularly scheduled updates rather than altering records already in the local catalog. UHL intends to proceed with this change irrespective of whether it is implemented in the cataloging copy provided by vendors, making the necessary updates through batch processes just as it has done in the past to clean up reproduction and born-digital records prior to load.

A more significant obstacle to full adoption of the provider-neutral standard at the local level is the lack of a reliable identifier to collocate equivalent manifestations on an automated basis. Two records for the same title from two different providers might share almost no metadata in common. Titles, edition statements, author entries, and ISBNs may vary in form and completeness. The ISBN comes closest to providing a standard identifier for electronic manifestations, but the ISBNs that appear in e-book records are too various to serve as an effective match point. UHL catalogers have seen nearly every possible type of ISBN attached to vendor-supplied e-book records: print, electronic, ten-digit, thirteen-digit, volume, and set. Data providers do not always include ISBNs, and many e-monographs have no ISBN assigned. The provider-neutral cataloging guidelines are predicated on a human cataloger comparing one digital file with another, or a digital file with cataloger-produced metadata, and not yet clear is how libraries that draw e-book records from many disparate sources can efficiently identify and merge duplicates.

Cataloging standards differ greatly between providers. Some e-content providers offer MARC delivery as an integral component of their online products and adhere closely to cataloging standards. Others are persuaded to offer MARC records only after ongoing negotiations with their library customers, providing skeletal records with no enhancements. Still others provide unusual enhancements that improve the richness and utility of the available metadata but are limited to a very small number of records. For example, one of UHL’s engineering e-book providers issues MARC records containing author, title, and subject access points at the chapter level. A provider of economics e-texts has made available records with minimal descriptive information and no name authority control, but rich abstracts and highly specific subject headings from a specialized thesaurus. As the provider-neutral model takes hold, catalogers must consider how to ensure that these rich access points are not lost when duplicates are merged in the local catalog. UHL can set load table parameters in UHL’s local catalog to preserve the contents of certain fields or groups of fields when a record is overlaid. UHL has used this approach in the past to protect critical metadata, but it requires access to and knowledge of integrated library system records load tables, expertise that is not universal in the library community.

The variation between content providers in the handling of multivolume sets is also a potential obstacle to provider neutrality. One provider might link to an entire multivolume set through a single URL and issue a single MARC record for that title. Another might link to each volume separately but bring all volumes together on a single title-level record. Still another might provide a separate record for each individual volume. Such disparate practices complicate the task of merging records, particularly if resources are identified by a volume title rather than the set title. UHL is presently content to let these different approaches coexist in its local catalog, but the question remains whether such records can or should be merged in the future. The optimal user experience for finding and obtaining access to multivolume sets is an area ripe for further study.

The current PCC guidelines encourage the use of classification numbers for e-monographs. UHL has always classified its electronic resources (e-resources) when cataloging individual titles, but classification numbers are not ubiquitous in vendor-supplied records. UHL retains any Library of Congress–type classification numbers that are part of the record set but neither verifies the correctness of existing classification numbers nor adds them to records that lack classification. When classification does exist, any existing shelflist number is removed from $b and replaced with “ebook.” The classification numbers are indexed in the local catalog and appear in the call number browse search, but they are not visible in the public record display because e-book records do not have attached item-level records (the means by which a call number displays to users in UHL’s catalog). Because e-book classification numbers are searchable but not publicly viewable, they serve primarily as a collection management tool rather than an access point.

UHL provides access to more than sixty e-book collections totaling nearly 400,000 titles and representing every type of e-monograph content mentioned above. Rarely does UHL catalog an e-book individually, but e-book management is nonetheless a resource-intensive process requiring strategy, compromise, and detailed documentation. Catalogers evaluate each package to determine the cataloging approach that will bring usable bibliographic data into the library catalog as quickly as possible with minimal cost for initial access and low overhead for ongoing maintenance.

Postcataloging maintenance and updates are an important part of e-book management. Each package has its own update schedule based on the nature of the package. Most of UHL’s large literary and primary source collections are growing slowly, and the providers periodically contribute new MARC records to the available record sets. UHL has found that, because of the sporadic and infrequent nature of changes, a yearly update is sufficient to keep these sets up to date. A full new record set is loaded once a year to overlay existing records and insert any new records that have been added to the set in the intervening year. Reference and academic e-book packages are more demanding because many of these sets have monthly additions and deletions. A monthly update is not itself particularly onerous, but juggling such updates for several providers and packages quickly adds up to a significant amount of record handling. For some reference and academic e-book collections, monthly record sets of additions and deletions are provided, but not for all. If the provider does not provide separate files for monthly additions and deletions, catalogers load an entire new set. Any existing record that is not overlaid during this process is presumed to have been dropped from the set and is deleted accordingly.

Vendor-supplied e-book records offer an efficient way of providing timely access to e-books, but good documentation is necessary to sustain the process. Most vendors do not provide express notification that new records are available, so the cataloger at UHL has the responsibility to seek out updates on a regular basis. Catalogers in libraries that subscribe to numerous e-book packages may find keeping current with the status and cataloging details difficult. These data could include unique ID prefixes, number of records in the last update, date of last update, update frequency, where to obtain records, and, if the records are behind an administrative login, how to gain access. The UHL cataloging department maintains a table in its departmental intranet space detailing the package name and provider, syntax for the unique identifier, date of last update, and review frequency. Although the records in the local catalog show the date of latest update, update schedules are tracked separately so that catalogers can see at a glance when to update any given collection.

Catalogers should have a way to identify, manipulate, and remove records in batch from a library system when maintenance is needed. Defining critical fields for record management (such as the 001 field, which groups records by provider and package, and the 910 field) to identify subsets within a larger group of records, has been crucial to achieving this exit strategy. When a change to an entire collection’s records is needed, catalogers can easily and reliably retrieve the entire batch for editing, output, or removal.

E-book record management has required UHL’s catalogers to cultivate a new awareness of the resource supply chain. More so than in the past, cataloging workflow decisions are closely connected to the manner in which a resource was purchased; a one-time purchase might be handled quite differently than a subscription to an aggregator, and an open access title differently than a major reference work. Cataloging concerns now have an opportunity to shape the direction of e-book purchasing. Curriculum and research support remain the primary criteria driving the acquisition of materials—as they should—but the needs of technical services can sometimes influence how materials are acquired. An unsustainable process is not beneficial to users, and UHL has begun to consider the total cost of resource management and access provision more closely as a factor in purchasing decisions.

E-books have made UHL’s cataloging managers more aware of organizational capacity. Batch record management requires a very different suite of skills from traditional cataloging. UHL has had to consider carefully how to acquire and allocate the specialized skills needed to perform this work, including load table management, large-scale data manipulation, independent problem-solving, and the rapid adoption of new tools and processes. UHL is still discovering how best to distribute this type of work to achieve the same systematic output and quality control that traditional cataloging processes now deliver. Cataloging managers at UHL consider the current processes too complicated and fragmented to be delegated with confidence. Finally, the move toward Web-scale services has forced UHL and other libraries to reconsider what has been sacrificed to take advantage of the economy and speed of vendor-provided e-book records. Tens of thousands of UHL’s e-resources do not show holdings in WorldCat, and holdings reclamation (reconciling non-OCLC local records with the WorldCat database and updating institutional holdings accordingly) is a complex and expensive process.

With few exceptions, UHL has found the MARC records supplied by e-book providers to be satisfactory, but three factors are moving the library away from using these records: the number of separate collections to be managed, the advent of individually purchased e-books, and the desire for provider-neutral records in UHL’s local catalog. While retrieving, editing, and loading record sets is straightforward, the sheer number of different providers and collections has made this process too cumbersome to continue in its present form. UHL cataloging managers are seeking a more streamlined process that can be delegated easily to staff. Individual e-book purchasing has raised the issue of how acquisitions personnel will communicate to cataloging personnel the availability of new e-books. Such communications can be easily lost or ignored, and scaling title-by-title notification up to large numbers of resources is difficult. Finally, provider neutrality has long been a goal for e-book records in the UHL catalog, a goal that cannot be achieved in the present environment unless the cataloging department devotes considerable effort to reviewing and deduplicating incoming records.

In fall 2009, UHL activated e-books within its e-resource knowledgebase (SerialsSolutions KnowledgeWorks) and began receiving e-book records. This approach is far from perfect—some of the records are derived from metadata in the knowledgebase rather than from cataloger-created MARC, and deduplication within the knowledgebase is an ongoing effort—but it satisfies the need for streamlined batch processing, efficient individual title processing, and provider neutrality. The transition to a single record provider will greatly reduce the number and variety of record loads that must be performed by the cataloging department, and because customization is applied to all e-book records coming from the service, records can be made to comply with the provider-neutral content standard without any further local editing. Regular notification and consistent records set the stage for a process that can be readily managed by paraprofessionals. As individual e-books are acquired, titles can be activated for public use by acquisitions staff as part of the receiving process, and the URLs can be verified in the knowledgebase at that time. A MARC record follows no more than thirty days later with the next monthly update. This process obviates the need for any kind of title-by-title communication to travel between acquisitions and cataloging, and the slight delay in making the MARC record available is offset by the fact that titles are available immediately through the e-resource portal. Finally, this approach enables UHL to rely on its vendor to do the strenuous and ongoing work of reconciling titles across providers and collections, work that—with the benefit of the vendor’s superior technology infrastructure and programmer support—can be accomplished much more readily by the vendor than by the library.

Over the last decade, batch loading vendor-supplied records has enabled UHL to provide access to e-books and other e-monographs efficiently and inexpensively without compromising other cataloging activities. As e-book collections have expanded and diversified, however, the profusion of platforms, service models, and metadata standards has strained UHL’s support infrastructure. UHL now uses a third-party resource management (MARC) service to reconcile titles across packages and supply provider-neutral records for as many resources and collections as possible. This approach has already given the library a simpler, more streamlined process that can be readily documented and delegated to cataloging staff, but it has not solved the problems inherent in mass record management. Until all of UHL’s e-book collections can be managed through a single service, catalogers must continue to download and manipulate records from multiple sources. Furthermore, relying on the services of a third party to improve local workflow and impose provider neutrality does not constitute a robust, lasting solution.

The Provider-Neutral E-Monograph Record Task Group was charged not only with developing a content standard for provider-neutral e-book records, but also with recommending “ways to promote the use of these records among … publishers/providers who create and issue cataloging copy for online monographic records” and “best practices for flexible use of these records.”²⁴ At the time of this writing (January 2010), none of UHL’s record providers had altered their records to adhere to the new guidelines. The provider-neutral record standard, if adopted widely, will lead to clearer and more consistent e-book records, so bringing data providers into compliance with the new guidelines will be an important first step toward improving access to e-books. However, in the absence of a robust identifier that could be used to match and merge e-book records from different sources, e-book records in the local catalog will continue to be provider-specific, even if each provider separately follows the provider-neutral content standard.

The authors are hopeful that the Task Group will continue to play a leadership role in pursuing dialogue with publishers and providers who issue cataloging copy. Promoting best practices for exposing titles, ISBNs, and other identifying information will help to better facilitate correct identification and duplicate detection for e-content, whether the work is done locally by a library or centrally by an e-content management vendor. In addition, it would be helpful for the Task Group to explore and document tools and best practices for batch processes, including efficient mechanisms for overlaying records, merging records, maintaining holdings for multiple providers, and automatically identifying records for which the last copy or version has been withdrawn. Mass management of bibliographic records is an activity that extends far beyond traditional cataloging into provider practices for exposing metadata, acquisition models, and the systems aspects of data management. Nonetheless, the effective provision and use of bibliographic records are essentially a cataloging problem. Moving beyond record creation standards to address best practices throughout the entire supply chain for e-book bibliographic data is the next crucial step that will enable libraries to provide clear, consistent, and timely access to e-books through their library catalogs.