Editors’ note: Today’s post is by Julie Zhu, Senior Manager, Discovery Partners at IEEE.

On August 26, 2026, NISO published KBART Phase III. The date took me back to the period soon after Phase I appeared in 2010, when I was a project manager for a publisher and served as its library liaison. When Jill Emery asked whether I could produce a title list for our publications, I searched online for a model, found the new KBART Recommended Practice, followed its format, and posted the result on our website.

Jason Price found the file. As a member of the KBART Working Group, he brought it to the group’s attention, and its chair contacted me with unexpected news: apparently, I had created the first publisher title list compliant with KBART Phase I. What began as a practical answer to one librarian’s question became my introduction to NISO and the start of more than 15 years with KBART.

KBART (Knowledge Bases and Related Tools) standardizes the title and coverage data that content providers send to knowledge bases, the systems libraries use to activate subscriptions and connect readers to content. My experience has taught me that a standard follows a recurring cycle: publication enables implementation; implementation exposes gaps; collaboration addresses them; and those discoveries shape the next standard. When longtime participant and current KBART co-chair Noah Levin encouraged me to tell this personal story, I realized that this cycle — not a sequence of specifications — was the story I wanted to tell.

A New World of Professional Service

That KBART-compliant title list — accidentally the first of many — led to an invitation to join the Phase II Working Group. Phase II was extending KBART beyond journals to monographs and conference proceedings, and I had firsthand experience with the variations in proceedings produced by 10 different publishers. I could bring practical examples and case studies to discussions of how the recommendations should accommodate them.

Joining Phase II was my first experience as a professional volunteer. Publishers, librarians, and library service providers brought different views of the information supply chain. No participant understood every part, but together we could see problems that crossed organizational boundaries.

Alongside their regular jobs, members examined fields, definitions, and exceptions with care. The process was democratic and sometimes slow because every perspective had to be heard and the result had to work across systems. That deliberation made the recommendations credible.

The experience also had personal significance. Having grown up in China and come to the United States for graduate study, I was still learning the culture of professional collaboration here. KBART’s outlook was international, and I felt included and respected. That group led me to NISO’s Open Discovery Initiative and other collaborations, beginning a pattern of volunteer service that continues today.

Phase II in Practice: Discovering the Gaps

When I became IEEE’s discovery service manager in 2014, KBART Phase II had just been published. I worked with cross-functional teams to automate title lists and improve the accuracy, completeness, and consistency of identifier, title, and coverage metadata. The files became more consistent, but automation did not end our discovery problems. Each week, I fielded dozens of library inquiries about missing titles, broken links, and which collections to select. I realized that an accurate publisher file could not ensure accurate representation and activation downstream.

A 2015 visit to a university library in London made the problem tangible. I found configuration issues affecting how subscribed collections appeared in its discovery service. Instead of treating every inquiry as an isolated case, I began analyzing where gaps originated. Some began with publisher metadata; others arose when providers loaded or indexed it; still others resulted from libraries selecting collections incorrectly. Often, the causes interacted.

Rachel Berrington, IEEE’s Director of Client Services, encouraged me to present this gap analysis at the Special Libraries Association Conference. Jalyn Kelley and I developed a poster. At the conference, Tony Stankus, then editor-in-chief of Science & Technology Libraries, stopped to examine it and invited us to develop it into an article. The resulting paper, “Collaborating to Reduce Content Gaps in Discovery,” quantified discovery gaps across 24 libraries that subscribed to the same IEEE collection but used four different discovery and link-resolver tools. It argued that publishers, discovery providers, and libraries could close those gaps only by treating discovery as a shared system. The article attracted several thousand downloads from readers in multiple countries, confirming that these were industrywide problems rather than isolated support cases.

The paper became a catalyst. EBSCO, ProQuest, and OCLC agreed to work with IEEE, and I visited their offices with IEEE Xplore leadership to meet with business and technical teams. Together, we investigated indexing, linking, metadata loading, and package representation — problems no organization could diagnose alone. One recurring question was measurement: How could a publisher know whether a provider had loaded a complete package? We worked on publisher-facing reports comparing KBART title counts with title counts in knowledge bases and record or item counts in discovery indexes. We also developed selection guides to help libraries activate the appropriate packages.

Working with librarians completed the picture. Through workshops, webinars, and consultations, I audited hundreds of library configurations. The reviews uncovered incorrect collection selections, overlapping activations, and local settings invisible to publishers and providers. Phase II implementation became a feedback loop: publisher metadata informed provider systems; provider reports exposed discrepancies; library configurations revealed implementation problems; and evidence guided corrections by all three stakeholders. KBART compliance described what should be discoverable. Auditing, measurement, and collaboration helped determine whether users could actually discover it.

Phase III: Preserving Titles and Identifying Packages

By Phase III, two forms of ambiguity had converged in my work: how to preserve the identity of a publication across languages and scripts, and how to preserve the identity of a package across systems.

The multilingual story began in East Asian libraries, where discovery systems did not adequately handle non-Latin scripts. Chinese, Japanese, and Korean text could display incorrectly, while transliterations and parallel titles disappeared. Between 2015 and 2018, members of the Council on East Asian Libraries compared knowledge-base metadata and saw recurring problems: abbreviated contributor names, inconsistent romanization, and missing parallel titles. These failures could make scholarship difficult to discover while stripping away part of its linguistic identity.

The limitations of Phase II’s first_author field were especially apparent. A surname alone is inadequate when names such as Wang are shared by millions. Libraries needed full contributor names and native, transliterated, and translated titles to coexist rather than replace one another.

In 2018, Nettie Lagace, then NISO’s Associate Director for Programs, invited me to speak about ODI and KBART at the CEAL China–U.S. Forum on Standards for Digital Resources. Afterward, Bie-Hwa Ma of the UC San Diego Library, who was helping lead CEAL’s metadata standards work, shared two central recommendations: include full names for Asian contributors, and support native, transliterated, and translated forms of titles. When Phase III work was underway around 2020, I brought these needs to the group.

Phase III translates those recommendations into practice through alternate_publication_title and alternate_title_type, which allow native, transliterated, and translated titles to coexist, and through contributor fields that support full names rather than surnames alone. What began as a regional discovery problem became part of a global standard.

Package identity presented a parallel challenge. IEEE is the case I know firsthand, but the underlying problems — unstable package names, mismatched identifiers, and incomplete downstream representation — are shared across content providers. As our number of packages and KBART files grew, it became difficult to track what had been delivered to each knowledge base. Collection names varied across systems, and packages with similar names could be confused. A library might activate the wrong collection or struggle to find its equivalent after changing providers.

I began building a master inventory of our KBART files: standardizing package names, gathering identifiers, and mapping packages downstream. During the Phase III work, longtime KBART leader and knowledge-base expert Benjamin Johnson proposed the name and initial format of the KBART Manifest. I immediately recognized its value. A manifest could describe a provider’s files, their purpose, and update schedules, reducing repeated communication.

I adapted the concept to our operational needs through the IEEE KBART Manifest. I established a manageable IEEE Xplore CollectionID for each package, standardized package names, and mapped each publisher identifier to the corresponding CollectionIDs in five major knowledge bases. This answered a recurring library question — Which collection should we select? — and made migrations between providers clearer without depending on names alone.

When a knowledge-base provider later asked us to identify forthcoming and recently released packages, I added Upcoming and New tabs. A package could appear before release, move to the main inventory when available, and remain visible as new during its early life. The Manifest became more than a directory; it documented a package lifecycle and supported earlier communication.

Years after our conversations about publisher-facing reports, I could obtain monthly title- and item-count data from providers. I added the counts to the Manifest and used formulas to compare our expectations with each provider’s report. A mismatch did not prove content was missing, but it signaled a need to investigate. Phase III brought multilingual representation, package identities, provider mappings, advance notice, and measurement into one operational story: make identities explicit, then use them to find gaps before libraries report them one by one.

The Next Connection: A Persistent Package ID

The Manifest makes packages easier to identify downstream, but it also exposes the next gap. Librarians have told us that a package needs an identifier that remains recognizable throughout its lifecycle. Names are unreliable: they change among publisher websites, quotations, invoices, knowledge bases, and library systems, and different packages may carry deceptively similar names.

I recently joined NISO’s Package ID Working Group, which began in May 2026 to explore a unique, persistent identifier for electronic-resource packages across the information supply chain. Ideally, the same identity would connect a package from its creation in a publisher catalog through quotations, orders, licenses, invoices, and entitlements; into KBART files; through subscription agents and aggregators; into discovery knowledge bases; and ultimately into usage reports.

The current IEEE KBART Manifest offers a working case study, but it solves only part of the problem. By mapping an IEEE Xplore CollectionID to collection identifiers in five knowledge bases, it connects equivalent identities downstream. The proposed Package ID aims to preserve one identity across the full lifecycle. Today, our CollectionID is not consistently linked upstream to an internal product code or carried through product-setup and commercial systems. A package on a quotation, invoice, or entitlement record may therefore remain disconnected from the package represented in KBART and discovery systems.

A persistent Package ID will not solve every metadata or configuration problem, and adoption across all stakeholders will take time. Yet, the direction is clear. KBART improved the exchange of package information downstream; the next challenge is to connect that exchange upstream, giving publishers, intermediaries, providers, and libraries a shared identity for the same product from creation through discovery, access, and assessment.

The Work Continues

My personal KBART journey has not been a straight path from one recommendation to the next. Each phase gained meaning through implementation: librarians articulated needs, publishers encountered gaps, providers revealed another part of the system, and collaboration turned shared experience into practice. Phase III shows how far that process has carried us — from a title list created almost by accident to multilingual discovery and a living map of packages across knowledge bases.

The lesson is transferable. A standard becomes infrastructure when organizations connect it to workflows, measure what happens after exchange, and return evidence to the community. The next stage needs broader participation, especially from publishers who can bring cases from product catalogs, sales, licensing, invoicing, and entitlement systems to the Package ID work. I encourage my colleagues and TSK readers to join NISO working groups and contribute the operational evidence on which durable standards depend. Standards may be published on a particular date; their value continues to be created together.