This past year, the Metropolitan Museum of Art in New York embarked on an ambitious plan to work with Wikimedia content, becoming one of the few GLAM institutions employing two Wikimedians within the institution to work on its open access initiatives.
Richard Knipel, Wikimedian in residence and Andrew Lih, Wikimedia strategist, explain how The Met is leading the way in comprehensive linked open data contributions (and continuing synchronization) of The Met's diverse collections with Wikidata and how this is spurring new experiments such as using artificial intelligence techniques in computer vision and machine learning to help generate depiction metadata. This is the first known use of machine learning on a large scale to contribute content to Wikidata. They will explain the history of The Met Museum collaboration with the Wikimedia community and the ongoing strategy for working with updating Wikidata and Structured Data on Commons. They will also describe the design and implementation of a new Wikidata Distributed Game that led to more than 3,000 decisions about artwork depiction statements.
Among the topics discussed include:
* In 2018-2019, the Metropolitan Museum of Art entered the third year of its open access initiatives.
* A quick history of The Met Museum's contributions and activities with the Wikimedia community (See: [[:en:User:Pharos/Wikimania]])
* A number of challenges in the course of the project that would be instructive for other GLAM entities:
** Defining a scope of work around two collections - The Met highlights (~2,000 items) and the Heilbrunn Timeline of Art History (~8,000 items).
** Crafting tools and scripts to perform detailed analysis and synchronization of Wikidata's partial collections information against an institution's database or API
** Addressing inadequacies in Wikidata modeling of materials, genres of artwork, dates, and other categories
** Determining best practices with mapping controlled vocabularies from an institution to Wikidata taxonomies, such as using custom terms versus well-accepted but unwieldy standards like Getty AAT.
** Formulating best collaborative practices around Wikidata tools, such as Distributed Game, Mix-n-match and other interactive tools
** Generating new co-creative demonstrations to showcase the benefits of open access collaboration, such as using Wikidata Query and multimedia tools
