Chief, Geography and Map Division Library of Congress
Learn how the Library of Congress staff is working to make 1 million records, with minimal level description in the form of handwritten folder labels, available for patrons to browse online and at their point of need through Machine Learning technology. The speaker will share the details of an experimental and ongoing project to test and compare three widely used and economical handwritten text recognition (HTR) tools for converting the handwritten folder labels into machine-readable text. Three machine learning API services are described and compared, as well as potential implications for large-scale transformation of handwritten metadata into fully searchable, machine-readable text will be explored.
Learning Objectives:
Understand the basic concepts of Machine Learning and how this technology is being used in libraries.
Learn how low-cost Hand Text Recognition Tools can be used to make library collections more accessible and discoverable.
Reflect on challenges and opportunities that arise from using Machine Learning in libraries