[B23] KuroNet: End-to-end Kuzushiji Transcription System for Understanding Historical Documents

Tarin CLANUWAT, Tarin CLANUWAT, Alex LAMB, Alex LAMB, Asanobu KITAMOTO, Asanobu KITAMOTO
2020 Dejitaru Akaibu Gakkaishi  
Kuzushiji was used as the dominant writing system in Japan for over a thousand years. However, following a reform of the Japanese language in 1900, Kuzushiji was removed from the regular school curriculum. As a result, most Japanese people today cannot read Kuzushiji. This causes difficulty in archiving and examining Kuzushiji documents. KuroNet is an end-toend Kuzushiji transcription system that transcribes Kuzushiji documents and takes one second per page. A service for running KuroNet is
more » ... lable online through IIIF Curation Viewer developed by the ROIS-DS Center for Open Data in the Humanities. In this paper, we explain the basic idea of KuroNet, such as the architecture and the capabilities of KuroNet. We also discuss how KuroNet can help museums and libraries preserve historical documents, making Kuzushiji documents more accessible, and how people can learn about Japanese culture and history from using KuroNet. Finally, we suggest directions for future work which could enhance the usage of Kuzushiji documents such as full-text search, text-to-speech, and other visualization projects.
doi:10.24506/jsda.4.2_165 fatcat:vkixvq4vwrb7rhixuqplzwmi3q