Librarians at University of Pennsylvania are using optical character recognition (OCR) to build a text corpus of Chinese-character documents from the early 20th century. YCW talks to Molly Des Jardin and Brian Vivier about their goals and challenges.