For over five decades, members of the Museum Computer Network (MCN) have assembled annually to discuss how technology impacts the work of Galleries, Libraries, Archives, and Museums—otherwise known as GLAM Institutions. Comprised of more than 600 institutions, the MCN promotes a collaborative spirit as the best approach towards achieving a broad range of digital initiatives, from building visitor engagement apps to publishing exhibition catalogues online to experimenting with artificial intelligence. Caitlin Sweeney, Director of Digital Publications, attended the 2019 conference in San Diego in early November to investigate where the conference’s conversation may overlap with the goals of the Wildenstein Plattner Institute.
One such topic was the consolidation and improvement of Metadata. To those who work with museum collection management systems, “metadata cleanup” is a very familiar topic. Why? Because the history of collecting metadata is extremely messy. Over the past few decades, many institutions have migrated their data numerous times as they upgraded databases; staff members came and went; researchers did not enter data according to agreed-upon standards; and CMS tools didn’t have sophisticated fields to capture information on a granular level. This means that lots of the data museums rely upon today is inconsistent, has lost its original context, and is difficult to navigate.
Increasingly, however, staff and visitors both want to conduct complex questions about museums’ collections. For example, you might want to know how many artworks in the collection were made by women, or whether a certain painting was shown in an exhibition in Tokyo in the 1980s, or if a museum has 19th-century photographs of Grecian ruins. In an era of search engines, asking these questions may seem like an easy task. Unfortunately, museums can only answer questions about the data they have in their CMS, and even then, only if it is structured correctly in the proper field.
In response to these demands, many institutions–including the WPI–are working to make their data searchable and machine-readable. On a practical level, this means consolidating duplicate material and using standardized names for institutions, people, and concepts (such as the Getty ULAN and ATT). On a political level, as museums commit themselves to social justice issues, curators, registrars, and metadata specialists are increasingly invested in creating metadata schema that address historical lacunae. For example, Chad Weinard from the Williams College Museum of Art pointed out that when the Guerrilla Girls were conducting their research, museums had never before been asked to supply the gender breakdown of the artists in their collection.
Now, this data is considered critical to understanding a collection, and many museums seek to rectify other absences in the data. For example, Jeremy Munro, from the Smithsonian National Museum of African Art, explained that museums ask the artists of new acquisitions to share their gender identity on intake forms. In turn, CMS tools can document a broader range of identities beyond the binary of “male” and “female.” As our understanding of diverse arts and artists becomes more sophisticated, museum data specialists have to think how to document this information in nuanced and complex ways. For example, Munro questioned how we should document the gender of artists who are no longer alive and cannot advocate for themselves, who belonged to a culture that embraced multiple genders, had been historically mis-gendered, or whose identity was never recorded at all. These issues demonstrate not only the challenges of improving metadata, but the historical significance of metadata stewardship.
This type of work is important to the WPI – especially if we strive to diversify our roster of artists. (You’ll have to wait until 2020 to learn about our new developments on that front!)
Metadata activism is exemplified by the work of our archivists, who are preparing a selection of our collection for public display in January. By identifying and tagging individuals–especially if in the past they might have only been known as a “Mrs.”–we are connecting them to a broader digital and scholarly arena, where they can be discovered, and where they can give a more robust impression of the time period.
Open Access data & its applications
Once relatively clean and structured, many museums have started to make their collection data available as downloadable datasets. As someone who has little experience working with digital raw data, my first question was why one want access to all this information? It can seem overwhelming.
At the MCN Datathon, Chad Weinard shared some amazing applications: Some people are making art. See, for example, this performance, A Sort by Joy (Thousands of Exhausted Things), by the Elevator Repair Service and the Office for Creative Research at the Museum of Modern Art in April, 2015.
Data visualization has enabled curators and researchers to ask different types of questions of a museum’s collection:
Artists are making strong statements with data visualizations–in Mona Chalabi’s case to show inequality in museums: https://www.theguardian.com/news/datablog/2019/may/21/museum-art-collections-study-very-male-very-white
Wikimedia-based initiatives, such as OpenArt, are using compiled and consistent metadata to enable people to browse images and make connections between works of art:
Andrew Lih shared one of the Met’s initiatives: datamaps that draw on wikidata to create event-based, non-hierarchical relationships between works of art and concepts. w.wiki/Aij
Jennie Choi, also from the Met, spoke about a collaboration with machine scientists at Cornell. They hosted a Kaggle competition and 520 teams participated (there wasn’t even a cash prize!) to build AI models that could automatically generate accurate metadata for images of artwork from the Met’s collection. You can learn more about the Met’s experiments with open access data here: https://www.metmuseum.org/blogs/now-at-the-met/2019/met-microsoft-mit-exploring-art-open-access-ai-whats-next
It turns out that there are lots of ways to use this data if you have a little imagination!
The cataloguing and archiving tool (CAT) that the WPI is developing is built to publish digital catalogue raisonnés and archival material. (More information on CAT here) Navigation of our data will be primarily through our publications and our search platform; users will be able to ask complex queries about the artists we focus on, as well as the associated archives, provenance, publications, and exhibitions. However, in the future, we hope that our data will not only be accessible in the ways we foresee. As artworks by particular artists are vetted, we hope to make our cataloguing data and images available open access (the latter of course dependent on copyright). By sharing clean data, we hope to contribute to valuable and diverse datasets, thus facilitating both niche research projects and wider initiatives in the digital humanities.