Including the public in digital catalogue descriptions will improve access to museum collections and help remove bias.
This article originally appeared in Museum magazine’s July/August 2024 issue, a benefit of AAM membership.
Over the past five years, conversations about biases in museums have migrated from internal staff discussions to stories in The New York Times and on the television show Last Week Tonight with John Oliver. While the public has been raising questions about whose history is preserved, who has input in their cultural legacy, where artifacts have been taken from, and more, the very definition of “museum” has changed.
After years of debate, in 2022, the International Council of Museums approved the following definition: “A museum is a not-for-profit, permanent institution in the service of society that researches, collects, conserves, interprets and exhibits tangible and intangible heritage. Open to the public, accessible and inclusive, museums foster diversity and sustainability. They operate and communicate ethically, professionally and with the participation of communities, offering varied experiences for education, enjoyment, reflection and knowledge sharing.” More succinctly, museums exist to engage the public in varied experiences, not just for study.
AAM’s “Excellence in DEAI Report,” published in 2022, has a similar emphasis on “the responsibility to tell all of our stories,” stating that “fostering space for each and every one of us requires intention, courage, and a commitment to the pursuit of excellence.” Museum professionals are not just encouraged but required by ethical guidelines to question the best practices in a field founded on colonialism in which rarities from around the world were sent to cultural centers in Europe and, later, the United States.
This questioning should extend beyond what is collected, what is displayed, and how it is displayed. One of the most basic, and most important, aspects of museum work remains largely obscured from the public: cataloguing. By improving the accessibility and searchability of their catalogues, museums can also address cataloguer bias and bring in a variety of user perspectives.
History of Cataloguing
In the 2020 publication Cataloguing Culture: Legacies of Colonialism in Museum Documentation, Hannah Turner tracks the history of cataloguing and access to museum collections. As early as the 1800s, catalogues using a single hierarchical classification structure were shaped into public displays and exhibitions that asserted neutrality while suppressing the complexity of cultural, social, and scientific systems.
Cataloguing began with curators creating lists and guides for use in the field during acquisitions, which were later enriched by cataloguing teams at the museum divorced from those collecting in the field. The language used in cataloguing and labeling objects emphasized attention to detail because these descriptions were often the only written record for an object. The descriptions were an attempt to standardize collections cataloguing, but Turner notes that by establishing consistent fields of information, certain information was valued over others.
Museums in the early 1900s continued to privilege physicality in catalogue descriptions. In 1904, the Field Museum of Natural History in Chicago had the following fields in their catalogue: when received; catalogue number; original or accession number; object; locality; number of specimens; received from; by gift, loan, or purchase; collected by; when collected; dimensions or weight; and remarks. More than a century later, the majority of museums still focus their cataloguing efforts on these fields.
As the second half of the 20th century dawned, computer use brought an increased awareness to the organization of catalogue information and its importance in facilitating research and access to collections. Due to technical difficulties, technology costs, and limitations on data storage and staff time, museums needed to determine the indispensable information catalogues would include. Different internal users required different information about objects, complicating the decisions of what to include in the catalogue.
In the early 2000s, these databases became online public access portals, effectively sharing collections with the public via the internet. This made it possible to search large amounts of data on a collection from anywhere in the world. But for the most part, these portals adopted the same strategies of the earliest card catalogues, resulting in the same types of information and biases of almost 200 years of description and cataloguing.
Where Museums Are Falling Short
Visitors to cultural heritage websites expect to navigate materials as they would on the broader internet, using thematic and contextual language. Most institutions use terms and descriptions that may be difficult for the public to understand, preventing meaningful access to collections and undermining the belief that collections are in fact for the public benefit, preserving history and culture for all. This disconnect not only limits access but also perpetuates polarization.
The internet is already a saturated knowledge space without a single narrative. By exclusively presenting professionally curated narratives online, museums fail to acknowledge the diversity of perspectives they embrace on-site: objects are interpreted differently based on the individual’s experience, and visitors engage with objects through various lenses.
In museums throughout the colonized world, collecting and preserving was often prioritized over sharing. This can still be seen in modern museums that publicly display 1–2 percent of their collection at any given time. According to a 2009 New York Times article, the Metropolitan Museum of Art in New York, for example, owned two million objects and displayed only tens of thousands at a time, and the Museum of Fine Arts in Boston had 18,000 objects on display out of an inventory of 450,000 works.
The public’s ability to discover collections materials relies on cultural heritage professionals’ anticipation of the language users will employ in their database queries. Directly transitioning catalogues meant to serve curators and staff to the web without considering the general public’s search needs and habits has created a semantic gap between what internet users can access and what they are looking for.
Descriptions often focus on physicality: who made the piece, where and when it was made, and its size and materials. Though helpful, this information does not aid a person looking for images that feature specific visual traits or thematic elements—for example, someone searching for snow or for Black history. Users want the descriptive metadata of objects to facilitate searches by keywords, subjects, and names.
What Museums Need to Do
For cultural heritage institutions to remain relevant and sustain public support, they must demonstrate that the objects held in public trust represent the collective memory of all. According to the January 2022 report Rethinking Relevance, Rebuilding Engagement: Findings from the Second Wave of a National Survey about Culture, Creativity, Community and the Arts, almost half (45 percent) of Americans participated in a community-based or participatory activity connected to a cultural heritage institution in the years before or during the pandemic, with little variation across race or ethnicity. The pandemic primed communities to expect cultural institutions to address social issues and offer online experiences that allow them to engage with collections and learning objectives. Doing so allows institutions to connect with communities and derive meaning from collections together.
Museums need to invite previously excluded voices into cultural heritage interpretations if they want to build and maintain public trust. Museums must understand how their digital offerings and footprints engender distrust. On-site, a gallery guide could explain a specific interpretation or steer a visitor toward other exhibitions for additional context. Online, there is no one to facilitate such an explanation if that information is not discoverable.
By prioritizing engagement as mutually beneficial, cultural institutions can develop projects that bridge the semantic gap while also bringing the mission-driven learning objectives of institutional programming to these projects’ design and goals. Museums that view community engagement as a process, rather than an outcome, can create online experiences as immersive as in-person ones.
Communities have already been invited into museums to serve as docents, or ambassadors to their cultures in the case of the Multaka project at Oxford University. Efforts like the Community Curation Project at the National Museum of African American History and Culture further invite the public into museum programming. Similarly, communities can provide invaluable insights into the cataloguing process.
Projects at the Metropolitan Museum of Art, San Francisco Museum of Modern Art, Adler Planetarium, Getty, and more have used crowdsourcing to augment museum catalogues, adding unique access points not found in traditional cataloguing. The Metropolitan Museum of Art’s tagging projects included an AI-run project and a collaboration with MIT and Wikipedia. The resulting Tagging Initiative now allows users to search more than 470,000 objects on subjects across media, cultures, and time periods. For example, it’s now possible to find 2,500 works across the Met’s extensive collection that feature dogs simply by searching on this term.
The Adler Planetarium’s Zooniverse.org-hosted crowdsourcing project similarly increased access to collections. The project attracted 6,976 individual participants to review more than 1,000 collections items, resulting in 322,993 individual metadata tags, 87 percent of which were completely new to the catalogue.
It is time for museums to change the way they fulfill their promise of preserving history and culture for all. Transparent processes that expand and enrich catalogue descriptions through community engagement can not only mitigate catalogue bias but also expose how language impacts daily internet usage. By fostering transparency and co-creation with the public, institutions can effectively represent community members in their collections and museums.
Sidebar
Can AI Help?
Machine learning and AI can enrich metadata and collections, but they also can perpetuate biases in the cataloguing process. AI tagging models like Google Cloud Vision API are trained on existing human-curated data sets that include conscious and unconscious biases in terms of race, gender, age, and emotion, embedding these biases in the algorithms from the onset.
In a 2019 Research Position Paper published by the global library organization OCLC, Thomas Padilla warned that the historic and contemporary biases create exactly this problem. Given that AI success in creating metadata and tags is still extremely dependent on the data set on which the AI model is trained, AI tagging is still inferior to the work of human participants, especially in terms of describing varied materials from multiple points of view.
For example, AI often fails to detect nuance or situational knowledge. For archival photographs at the Adler Planetarium, Google Cloud Vision API inaccurately tagged scientists in lab coats as doctors or nurses. Human volunteers added tags for “scientists,” “labcoats,” and “lab coats” but did not include tags for “doctor” or “nurse” as the images did not show medical equipment or hospital backgrounds that would prompt such tags. However, the Adler’s inclusion of AI tags in its Zooniverse.org project clearly enticed and motivated some volunteers’ participation: the workflow that featured AI consistently saw two to three times more engagement than the non-AI workflow, demonstrating the appeal of AI, automation, and algorithms.
Museums should not consider AI as a replacement for metadata crowdsourcing projects or professional cataloguing but as a motivation and incentive for public participation in these projects.
Pull Quote
“Directly transitioning catalogues meant to serve curators and staff to the web without considering the general public’s search needs and habits has created a semantic gap between what internet users can access and what they are looking for.”
Resources
Hannah Turner, Cataloguing Culture: Legacies of Colonialism in Museum Documentation, 2020
“Rethinking Relevance, Rebuilding Engagement: Findings from the Second Wave of a National Survey about Culture, Creativity, Community and the Arts,” Slover Linett, January 31, 2022
sloverlinett.com/insights/rethinking-relevance-rebuilding-engagement-findings-from-the-second-wave-of-a-national-survey-about-culture-creativity-community-and-the-arts/
Thomas Padilla, “Responsible Operations: Data Science, Machine Learning, and AI in Libraries,” OCLC Research Position Paper, 2019
doi.org/10.25333/xk7z-9g97
Comments