@techreport{oai:ipsj.ixsq.nii.ac.jp:00191438, author = {Karol, Nowakowski and Michal, Ptaszynski and Fumito, Masui and Karol, Nowakowski and Michal, Ptaszynski and Fumito, Masui}, issue = {2}, month = {Sep}, note = {Ainu is an endangered language of the Ainu people, native inhabitants of northern Japan. It has been the subject of many studies, but most scholars work on small amounts of language data. We propose a corpus of Ainu covering a wide range of documents, in a consistent structure that will enable large-scale linguistic analysis and support the development of NLP technologies for Ainu, contributing to the process of its revitalization. The corpus contains parallel text in Ainu and Japanese. Its subset includes POS annotations produced by Ainu language experts. For the remaining parts, annotations will be generated automatically. At present, resources collected for the corpus comprise 1.86M characters (410K tokens) of text in Ainu. Their utility for NLP applications has been verified by applying them in a tokenization system, which achieved average F-score above 95%., Ainu is an endangered language of the Ainu people, native inhabitants of northern Japan. It has been the subject of many studies, but most scholars work on small amounts of language data. We propose a corpus of Ainu covering a wide range of documents, in a consistent structure that will enable large-scale linguistic analysis and support the development of NLP technologies for Ainu, contributing to the process of its revitalization. The corpus contains parallel text in Ainu and Japanese. Its subset includes POS annotations produced by Ainu language experts. For the remaining parts, annotations will be generated automatically. At present, resources collected for the corpus comprise 1.86M characters (410K tokens) of text in Ainu. Their utility for NLP applications has been verified by applying them in a tokenization system, which achieved average F-score above 95%.}, title = {A proposal for a unified corpus of the Ainu language}, year = {2018} }