OCR correction by improving robustness to orthographic variation in yiddish, a low resource language. [this proposal will support the development of an experimental system to improve error…

The National Endowment for the Humanities (NEH) awarded a $138,319 Digital Humanities Advancement Grant (CFDA 45.169) to the National Yiddish Book Center, Inc. to develop an experimental system for improving optical character recognition (OCR) error correction on its extensive corpus of digitized Yiddish texts. This 2-year project, beginning February 1, 2024, aims to enhance the accessibility and usability of the Center's digital collection, which is expected to double to over 12,000 digitized Yiddish books by 2025 through the ongoing Universal Yiddish Library project. The experimental OCR correction system will be applied to the Center's existing 650 million-word corpus of digitized Yiddish text, improving the accuracy and quality of this invaluable cultural heritage resource for researchers and the public.

OCR correction by improving robustness to orthographic variation in yiddish, a low resource language. [this proposal will support the development of an experimental system to improve error correction on the existing yiddish book center optical character recognition (OCR) projects, which are the first and largest instances of OCR for yiddish done at scale. The experimental systems will correct a 650 million word corpus (about 12,000 scanned yiddish books) of yiddish text produced by optical character recognition (OCR) which is available now at the yiddish book center?S full-text search website (http://OCR.yiddishbookcenter.ORG). This corpus will double in size by 2025 upon the completion of a separate OCR project, the universal yiddish library. The universal yiddish library project will add at least 10,000 additional books from multiple institutions to the existing corpus, and the OCR correction methods used for this project will be applied to the universal yiddish library corpus in the future.]