Rhys Owen & Andrew McGhie- Wrestling with Qilin-the Challenges of Chinese OCR

About

Wai-te-ata Press is custodian of the only collection of classical Chinese metal types in New Zealand. Imported from Hong Kong, these types were used to print the NZ Chinese Growers Monthly Journal from 1952-1972. Auckland Libraries in collaboration with the Chinese Heritage Poll Tax Trust digitised these newspapers which now form the basis of an innovative digital humanities research project at Victoria University of Wellington. By investigating new generation approaches to Chinese OCR through image segmentation and type edit distance, we are developing a blended, at-scale platform and workflow that links our language expert back to the physical types she is cleaning, classifying, and re-housing while providing an experimental locus for addressing fundamental issues of OCR quality. In the process, this moves our engineering students forward into the very real challenges of machine learning, neural nets and machine translation, all in a language they do not or need not understand.

Andrew McGhie - Victoria University of Wellington

Rhys Owen , Wai-te-ata Press : Te Whare Ta O Wai-te-ata Victoria University of Wellington , Technical Lead

Archive

Wikimedia Commons, in Wikimedia Commons

Licence or status

CC BY 3.0, as the Wikimedia Commons states it.

Rights statement

Creative Commons Attribution 3.0

Credit

National Digital Forum

Subtitles

English, Chinese

Who can watch

The archive gives no rating. Invisible Network shows it to Teens profiles and older, judged from its collection and subjects. How films are placed