@inproceedings{saunders-etal-2020-inference,
    title = "Inference-only sub-character decomposition improves translation of unseen logographic characters",
    author = "Saunders, Danielle  and
      Feely, Weston  and
      Byrne, Bill",
    editor = "Nakazawa, Toshiaki  and
      Nakayama, Hideki  and
      Ding, Chenchen  and
      Dabre, Raj  and
      Kunchukuttan, Anoop  and
      Pa, Win Pa  and
      Bojar, Ond{\v{r}}ej  and
      Parida, Shantipriya  and
      Goto, Isao  and
      Mino, Hidaya  and
      Manabe, Hiroshi  and
      Sudoh, Katsuhito  and
      Kurohashi, Sadao  and
      Bhattacharyya, Pushpak",
    booktitle = "Proceedings of the 7th Workshop on Asian Translation",
    month = dec,
    year = "2020",
    address = "Suzhou, China",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingest_wac_2008/2020.wat-1.21/",
    doi = "10.18653/v1/2020.wat-1.21",
    pages = "170--177",
    abstract = "Neural Machine Translation (NMT) on logographic source languages struggles when translating {\textquoteleft}unseen' characters, which never appear in the training data. One possible approach to this problem uses sub-character decomposition for training and test sentences. However, this approach involves complete retraining, and its effectiveness for unseen character translation to non-logographic languages has not been fully explored. We investigate existing ideograph-based sub-character decomposition approaches for Chinese-to-English and Japanese-to-English NMT, for both high-resource and low-resource domains. For each language pair and domain we construct a test set where all source sentences contain at least one unseen logographic character. We find that complete sub-character decomposition often harms unseen character translation, and gives inconsistent results generally. We offer a simple alternative based on decomposition before inference for unseen characters only. Our approach allows flexible application, achieving translation adequacy improvements and requiring no additional models or training."
}