The Library of Congress is looking to build on experimental processes that expand digital access to the library’s collections by combining a human component to its computational methods – known as human-in-the-loop approaches.
“Over the past few years, the Digital Innovation Labs Section (Labs) of the Digital Strategy Division has undertaken a range of programs aimed at maximizing the use of digital collections and supporting emerging research methods,” a request for information (RFI) says. “Through various experiments, reports and events, Labs has been committed to exploring methods and approaches that responsibly integrate people and machines.”
Through a contractor, the library wants “at least two experimental prototypes or proofs of concept for at least two human-in-the-loop workflows using Library of Congress collections that are presented and tested with users.”
These prototypes are meant to model, test, and evaluate different ethical approaches to utilizing crowdsourcing and machine learning methods that will enhance the usability, utility, discoverability, and user engagement of the library’s digital collections.
“At least one experimental workflow shall start with machine learning derived metadata [and] create a prototype crowdsourcing task and interface that invites users to add feedback on the accuracy of the derived metadata,” the RFI states. “The primary use of the publicly derived metadata shall be to serve as training data for a machine learning algorithm that shall be applied to Library digital collections to create enhanced metadata.”
Submissions by interested parties are due Aug. 5, 2020.