ABSTRACT
This session discussed the knowledge base metadata lifecycle, current and upcoming metadata standards, and the role that knowledge bases play in e-resource management. The session provided different perspectives on knowledge bases, including from librarians, one of whom is a former knowledge base e-content analyst, and product managers, as well as a discussion of the National Information Standards Organization (NISO) Knowledge Bases and Related Tools (KBART) Automation: Automated Retrieval of Customer Electronic Holdings recommended practice and what this could mean for knowledge bases in the future. The session included a conversation regarding how leveraging knowledge bases can aid librarians in improving resource discovery within their own libraries and ultimately decrease the amount of time spent on metadata workflows.
KEYWORDS:
One of the difficulties with understanding knowledge bases and how they are leveraged is that librarians generally are interacting with a product that utilizes a knowledge base rather than the knowledge base itself. Vendors such as EBSCO, ProQuest, and OCLC build services and solutions that integrate with their knowledge bases, such as Publication Finder (EBSCO), 360Link (ProQuest), and OCLC Link Resolver. Generally, if a librarian utilizes a product from vendor such as EBSCO, ProQuest, or OCLC WorldShare, they may be benefiting from a knowledge base, but they are not interacting with the raw data in the knowledge base.
A knowledge base serves several purposes. First, a knowledge base brings together a vendor’s grouping of title-level metadata organized by the collection in which the title can be acquired. It also brings together publisher title lists with the goal of further grouping together identical or related resources to aid in the identification of the appropriate copy. Alone, a knowledge base is simply a grouping of metadata, rather than a discovery tool used by libraries. However, library vendors develop tools that utilize the knowledge base’s vast data. These tools are used by library employees for tasks such as electronic resource management and they are also used by library patrons to locate resources. The purpose of these tools is to allow for seamless discovery for the end user. Ultimately, a knowledge base supports discovery, but it is not a discovery layer unto itself.
To increase the discovery of library resources, many libraries are already benefitting from a knowledge base, but it may not be as integrated into integrated library systems (ILS) as they would like it to be. Traditional ILSs place the bibliographic database and the patron database as the backbone of the system.Footnote1 If knowledge base data is included within the ILS, rather than holding a centralized place, it is often utilized as an add-on module or entirely external to the ILS structure. As vendors have moved towards providing new services and technologies, the knowledge base is becoming more central to the structure of metadata organization.
However, when we look at the structure of next generation library platforms like the Future of Libraries is Open (FOLIO), Ex Libris’ Alma, and OCLC’s WorldShare Management Services, the knowledge bases are more central to their structure, as knowledge base data often serves to link important bibliographic, holdings, and access information in one place.Footnote2 With this structure and further Knowledge Bases and Related Tools (KBART) AutomationFootnote3 implementation, it is possible to envision a world where not only entitlements will be communicated through KBART data, but also license terms and business terms like perpetual access or interlibrary loan permissions. Next generation library platforms and the knowledge bases that power them have the potential to support transfers of electronic data interchange (EDI) and KBART data all at once to populate our orders, entitlements and all of the associated terms.
To understand how to best leverage knowledge base information in individual library systems, it is important to understand how data is organized. As previously stated, a knowledge base is a vendor’s grouping of title-level metadata organized by collection. When that metadata is combined with metadata from as many vendors as possible, we now have, essentially, as complete of a record of titles as possible. Of course, having a record of everything is impossible, for a number of reasons, including that titles are published constantly, all over the world. The metadata needed to make collections discoverable generally comes in two different formats: MARC records and KBART files. Generally, libraries most often use MARC records, and knowledge bases most often use KBART files. Unlike MARC records, KBART files are focused on the entire collection as a whole and are less focused on the metadata attributes of an individual resource.
KBART, which stands for Knowledge Bases and Related Tools, is a NISO initiative, which first published KBART Phase I in 2010.Footnote4 KBART Phase I primarily focused on communicating information about journals.Footnote5 KBART Phase II was established in 2014, and it introduced new fields to include more data about books and conference proceedings.Footnote6 As of this writing, NISO has initiated the development of KBART Phase III, and is working towards making additional revisions, possibly to add additional resource types.Footnote7 Notably, in June 2019, NISO introduced KBART Automation.Footnote8
Collections, or packages as they are sometimes known, come in a variety of different formats. Some vendors or publishers may sell multiple collections with multiple titles, while other vendors may choose to sell titles one at a time. Regardless, a vendor may produce a list of titles included in that collection, which includes identifiers and other important pieces of information.
A library may have a different perspective of the data in a knowledge base because the library aims to keep a record of everything that the library may access or own, not everything that is available to access or own. For example, one library may decide that they want to purchase two titles from a collection rather than the entire collection.
EBSCO Information Services, as a knowledge base provider, must view its knowledge base from two perspectives: as a repository of global data and as a service provider to individual libraries. The EBSCO Knowledge Base contains global data, including title lists, packages, and embargos, which is the traditional view of what a knowledge base is, but this information is also managed for libraries by capturing and reflecting every single customer’s unique holdings against the global data. Because of this, there is a dual task of supporting the global data, but also of supporting the data that is unique to each customer. As of the date of this presentation, EBSCO is one of 1,900 vendors supplying content to the EBSCO Knowledge Base and of the 15,000 packages available, 647 are provided by EBSCO. The volume of global data, combined with the unique data for each customer, is significant.
From a vendor perspective, the role of the knowledge base is changing from how we think traditionally about the knowledge base. Holdings management, publication browse, and some attachment to the discover layer all remain a part of the knowledge base. However, the knowledge base is beginning to power more of the e-resource management lifecycle, and it is becoming an engine around which the library can simplify workflows and de-silo data. As the heart of the resource management platform, the knowledge base can help other parts of the e-resource lifecycle function, including integrated APIs that can move information to additional interfaces. From the beginning, knowledge bases powered key discovery tools like link resolvers; now with the layering of additional data like cost and usage, knowledge bases support usage analysis and collection management in brand new ways. The knowledge base is the hub of the e-resources lifecycle, which is a cyclical process and requires review and modification. It is expected that the industry, as a whole, will continue to see the knowledge base as an engine to support new aspects of the e-resource lifecycle.
The EBSCO Knowledge Base processes large quantities of data every day. The growing content is managed with dedicated staff working to ensure that the knowledge base is fully updated and that it supports any and all content that libraries want to add to their holdings. In addition to the sheer volume and scale of the data, knowledge base vendors are thinking about how to best respond to changes in the data as well as assisting customers in maintaining their own custom content in as efficient a manner as possible. However, in order to develop in the direction that continues to support library workflow needs both now and in the future, feedback from knowledge base users is key. Understanding the specific workflows or issues that library staff face every day and the challenges that library workers would like to see a knowledge base help solve is essential to developing vendor knowledge bases in a way that benefits everyone in the future.
To build that next generation knowledge base, key stakeholders must first acknowledge the gaps in knowledge base performance that libraries must mitigate on a daily basis. To begin with, knowledge base data has three stakeholders: the publisher, the knowledge base vendor, and the library. This can create a difficult situation when it comes to disagreements regarding metadata. The publisher provides metadata to the knowledge base vendor, the library disagrees with the accuracy of the publisher’s metadata, and the knowledge base vendor is stuck in the middle. Between title changes, variations in completeness of coverage, and variations in entitlements that can occur at the resource level, metadata discrepancies can be daunting to resolve for librarians who are attempting to identify their holdings at the collection level. These metadata disagreements can be a symptom of a larger problem in this three-stakeholder system, a lack of effective and standardized communication between librarians, publishers and knowledgebase vendors. Effective communication is not only key when sharing metadata, it also important that all three stakeholders are accurately representing how resources are purchased and bundled within their knowledge base.to ensure that all parties are on the same page about a libraries’ entitlements.
Another persistent problem when it comes to knowledge bases and the library services they support is the appropriate copy problem. Getting the user to the appropriate copy of the resource they are seeking promises to remain an issue as long as libraries have a hybrid environment of access that includes aggregated access, subscribed access, pay-per-view access, Demand Driven Acquisition (DDA) and Evidence Based Acquisition (EBA) access, Open Access (OA) and many more models. What may be the appropriate copy for a user in one library may not be for another, and the knowledge base must be able to support this variation and prioritization on a very granular level. Libraries and their knowledge bases are also faced with the problem that many of their users are doing their research and finding content entirely outside of library platforms and affiliations which means there is not always a system to direct traffic and send the user to the most appropriate copy. Appropriate copy promises to be even more important as we begin to see more and more content published and assigned Digital Object Identifiers (DOI), like data sets and manuscripts that previously were considered unpublished or ancillary content.
Libraries need the ability to track and manage the rights associated with the packages and titles they activate in their knowledge base. Historically, this process has been a part of electronic resource management systems, which are not universally a component of standard ILSs. A library’s ability to track how an e-resource was acquired, what content they are entitled to access, whether they are entitled to perpetual access and, if so, how much and what the purchased access will look like in a year, ten years and fifty years from now are key to the management of e-resources. When titles are selected in library facing knowledge base systems, such as WorldShare Collection Manager or EBSCOadmin Holdings Management, this information is not typically ascertainable from knowledge base data. Electronic resource licensing and in some instances access data are examples of information that is specific to the individual subscribing institution while also being tied to a specific resource, unlike data such as title or DOI that would be universal. Rights like text and data mining or interlibrary loan are key to determining how a library will use a resource and the tracking of this data is key to automating the discovery of these resources for users as well as other library units like resource sharing.
All of these challenges feed into probably the biggest challenge of knowledge bases for librarians, which is their set up and maintenance. Because a knowledge base comes to librarians with all the world’s resources, libraries must take the staff time to customize and continue to refine the knowledge base to their own resources. This can be a formidable task even when a library purchases mostly aggregated content because these resources are constantly changing. This constant change can include maintaining changes to Open Access materials and supporting trials and time-limited free offers like those made available during the COVID-19 pandemic. While libraries are excited to have more access to content for their users, making that content available to users in a timely fashion is time-consuming. While the customization of resources is often unique to each library, addressing errors in metadata and changes to resources should not be. However, there can be a lot of duplication of effort on the part of individual libraries to address problems that are systemic rather than unique to the library.
At the North Carolina State University Libraries, the gaps in the knowledge bases are addressed through a process called package management. This process includes having a staff member designated as a package manager who uses a custom database to reconcile order data against entitlement title lists from the publisher. Any data that does not match gets passed to staff members who act as package owners for the libraries’ packages. These package owners investigate discrepancies and make corrections. Once the publisher entitlements and order data match, this data is then reconciled against the title lists from our knowledge base to ensure that these access points are properly customized. Any data that does not match gets passed to that package’s owner for investigation and correction. This can be a lengthy process when the data does not match, but going through a process such as this is necessary to ensure that entitlements are what they should be to support user access.
While this process addresses entitlements in our discovery systems, it does not address errors in entitlements that occur once the user is on the publisher’s site. This is where access verification comes in. Package owners identify packages that may be vulnerable to access issues. This can include platforms that have had access issues in the past or have recently undergone a platform migration or entitlement restructuring, like going from a “Big Deal” to individual subscriptions. For all other packages, the library takes a fix at failure approach where we address access disruptions as they are identified by users. While neither of these approaches is foolproof as they do not catch all problems, they serve to create some data quality measures.
KBART Automation could assist with these gaps by encouraging more direct communication between publisher systems and user systems to populate entitlements and access points. Better integration between knowledge base data and bibliographic and order data from the ILS, usage data, and license rights and limitations can also allow libraries to evaluate collections more holistically.
KBART Automation includes concepts that could help solve problems in the metadata lifecycle. Libraries could have faster delivery of metadata across their systems, duplicate efforts across systems could be reduced, and holdings information could be more accurate. If each library’s holdings information was supplied directly from the publisher who sold the resource, then reasonably, we can assume that the holdings information for the library will be more accurate. The librarian would no longer have to select individual titles from a collection of many titles and could, in effect, bypass the traditional interaction with the knowledge base. KBART Automation also could create new problems, namely that problems in metadata would be less visible. There could also be potentially less awareness of when updates are made and fewer opportunities for customization for librarians. Automation, with its many benefits, also has downsides that could negatively impact discovery. Librarians would have fewer opportunities to review and identify any discrepancies in data. New workflows may need to be implemented to allow the librarian to detect and resolve these discrepancies.
As the knowledge base’s role changes beyond linking management and moves into new avenues, such as usage management, there is additional pressure placed on knowledge base development. Workflows in libraries are evolving and so are the places where decisions are being made, so vendors are incentivized to present the data where it will most benefit the end user. For example, a vendor might present usage data closer to the library’s holdings data, while also bringing that together with the library’s ILS. Ultimately, the goal for knowledge base vendors is to provide the right data at the time of service, both to the librarians and to the end user. In many circumstances, too much data can be as problematic as not enough. Library staff need the ability to drill down and see the data to make a purchase decision or to ensure that the right copy is turned on or that the link is configured in the right order. For vendors, this requires feedback and conversations with users of the knowledge base to understand their pain points and evolving data needs. Vendors are interested in solving user problems, and by having these ongoing conversations about the role knowledge bases are playing in the library, knowledge bases can continue to develop to meet those new needs.
These improvements will not necessarily mean that libraries will no longer perform activities like reconciliation and access verification, but it may mean that the errors are fewer and libraries can focus more on reaching users wherever they are rather than on making sure that entitlements match across the publisher, knowledge base vendor, and library.
Further reading
Nettie Lagace, “The NISO KBART Automation Working Group: Modernizing Knowledgebase Communications,” Serials Review 44, no. 3 (2018): 251–2. doi: 10.1080/00987913.2018.1522200.
Disclosure statement
No potential conflict of interest was reported by the authors.
Additional information
Notes on contributors
Elizabeth Levkoff Derouchie
Elizabeth Levkoff Derouchie is Assistant Librarian and Metadata Librarian for Serials & Electronic Resources, Samford University, Birmingham, Alabama.
Beth Ashmore
Beth Ashmore is Associate Head, Acquisitions & Discovery (Serials), North Carolina State University, Raleigh, North Carolina.
Eric Van Gorden
Eric Van Gorden was Senior Product Manager, EBSCO Information Services.
Notes
1 Karen Newbery, “The Backbone of the Library, the Library Catalog,” Bitstreams: News From the Digital Projects Team (blog), April 12, 2018, https://blogs.library.duke.edu/bitstreams/2018/04/12/the-backbone-of-the-library-the-library-catalog/ (accessed December 17, 2020).
2 Carl Grant, “The Future of Library Systems: Library Services Platforms,” Information Standards Quarterly 24, no. 4 (Fall 2012): 4–15, https://doi.org/10.3789/isqv24n4.2012.02.
3 “NISO RP-26-2019, KBART Automation: Automated Retrieval of Customer Electronic Holdings,” NISO.org, June 18, 2019, https://www.niso.org/publications/niso-rp-26-2019-kbart-automation (accessed December 17, 2020).
4 “Knowledge Bases and Related Tools (KBART),” NISO.org, https://www.niso.org/standards-committees/kbart (accessed August 20, 2020).
5 Ibid.
6 NISO, “Knowledge Bases and Related Tools”; and “NISO RP-9-2014, KBART: Knowledge Bases and Related Tools Recommended Practice,” NISO.org, March 27, 2014, https://www.niso.org/publications/rp-9-2014-kbart (accessed August 20, 2020).
7 NISO, “NISO RP-26-2019.”
8 NISO, “Knowledge Bases and Related Tools.”