William & Mary awarded grant to support digital humanities project

Students on computers
Connor Fenton ’17, Dan Delmonaco ’16 and Emily Zinger ’17 transcribe royal documents in Swem Library’s digital lab in spring 2017.

William & Mary has been awarded a grant by the National Endowment for the Humanities (NEH) to support the advancement of digital humanities. The grant will fund a collaborative project between the university’s libraries and computer science department to improve the capabilities of Transkribus, a handwritten text recognition platform, to process tabular data.

Library staff and student assistants currently use Transkribus to transcribe materials for the Georgian Papers Programme, an international digital humanities project with the Royal Archives in the United Kingdom and King’s College London to make available online historic manuscripts relating to the Georgian monarchy.

“We are thrilled to receive a National Endowment for the Humanities grant, and are eager to work with Dr. Zhenming Liu in Computer Science on automated handwritten text recognition for the Georgian Papers Programme,” said Debbie Cornell, head of digital services for the libraries.

The GPP launched at Windsor Castle in April 2015. That fall, W&M's Reves Center facilitated a partnership between the Royal Archives and King's College London and W&M Libraries and the Omohundro Institute of Early American History & Culture.

Transkribus eliminates having to manually transcribe thousands of pages of handwritten materials. Instead small sets of documents are transcribed and used to inform or “train” the tool to decipher handwriting and build a model to produce transcriptions.

While Transkribus is incredibly useful, it does have limitations.

diaries
Diaries belonging to Queen Charlotte, consort of George III. (Royal Archives)

“Though we have worked with Transkribus to develop a model to read the script of George III, the platform has difficulties with handling complex tabular data, where text is not formatted in consistent grid-like formats,” said Cornell. “We, like other Transkribus users, would benefit from improvement in functionality to recognize and export manuscript texts, such as inventories, accounts, and ledgers to spreadsheets.”

The $100,000 grant will support development of new machine learning techniques to enhance Transkribus’ ability to recognize text in tabular form and to share new developments with other Transkribus users.

The NEH today announced $29 million in awards for 215 humanities projects across the country. This round of funding, NEH’s third and last for fiscal year 2019, will support vital research, education, preservation and public programs in the humanities.

“NEH grants help strengthen and sustain American cultural life, in communities, at museums, libraries, and historic sites, and in classrooms,” said NEH Chairman Jon Parrish Peede in the announcement. “As the nation prepares to commemorate its 250th anniversary in 2026, NEH is proud to help lay the foundations for public engagement with America’s past by funding projects that safeguard cultural heritage and advance our understanding of the events, ideas, and people that have shaped our nation.”