The DOI matches its Crossref record; 20 of the 21 references it cites were found in a registry or an index.
Neural Machine Translation of Logographic Language Using Sub-character Level Information
Open accessAbstract
Recent neural machine translation (NMT) systems have been greatly improved by encoder-decoder models with attention mechanisms and sub-word units.However, important differences between languages with logographic and alphabetic writing systems have long been overlooked.This study focuses on these differences and uses a simple approach to improve the performance of NMT systems utilizing decomposed sub-character level information for logographic languages.Our results indicate that our approach not only improves the translation capabilities of NMT systems between Chinese and English, but also further improves NMT systems between Chinese and Japanese, because it utilizes the shared information brought by similar sub-character units.
Overview
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, L., & Polosukhin, I. (2017) · arXiv · doi:10.48550/arxiv.1706.03762 · Cited by 120,404
Jakubův, J., Chvalovský, K., Goertzel, Z., et al. (2023) · DROPS (Schloss Dagstuhl – Leibniz Center for Informatics) · doi:10.4230/lipics.itp.2023.19 · Cited by 76,648
Learning Phrase Representations using RNN Encoder–Decoder for Statistical Machine Translation
Cho, K., van Merrienboer, B., Gulcehre, C., Bahdanau, D., Bougares, F., Schwenk, H., & Bengio, Y. (2014) · Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.3115/v1/d14-1179 · Cited by 25,545
Sequence to Sequence Learning with Neural Networks
Sutskever, I., Vinyals, O., & Le, Q. V. (2014) · arXiv · doi:10.48550/arxiv.1409.3215 · Cited by 17,766
Neural Machine Translation by Jointly Learning to Align and Translate
Bahdanau, D., Cho, K., & Bengio, Y. (2014) · arXiv · doi:10.48550/arxiv.1409.0473 · Cited by 16,213
Effective Approaches to Attention-based Neural Machine Translation
Luong, T., Pham, H., & Manning, C. D. (2015) · Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing · doi:10.18653/v1/d15-1166 · Cited by 8,382
Google's Neural Machine Translation System: Bridging the Gap between Human and Machine Translation
Wu, Y., Schuster, M., Chen, Z., Le, Q. V., Norouzi, M., Macherey, W., Krikun, M., Cao, Y., Gao, Q., Macherey, K., et al. (2016) · arXiv · doi:10.48550/arxiv.1609.08144 · Cited by 5,319
Statistical Significance Tests for Machine Translation Evaluation.
Koehn, P. (2004) · Cited by 1,455
Subword Regularization: Improving Neural Network Translation Models with Multiple Subword Candidates
Kudo, T. (2018) · Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · doi:10.18653/v1/p18-1007 · Cited by 1,021
Statistical significance tests.
Cox, D. (1982) · British Journal of Clinical Pharmacology · doi:10.1111/j.1365-2125.1982.tb01987.x · Cited by 335
Research works printed but not linked: 2
- Sepp Hochreiter and Jürgen Schmidhuber. 1997. Long short-term memory. Neural computation, 9(8):1735-1780.
- Rico Sennrich, Barry Haddow, and Alexandra Birch. 2016. Neural machine translation of rare words with subword units. In ACL 2016, pages 1715-1725.
Cited by
Sorted by citations
Koehn, P. (2020) · Cambridge University Press eBooks · doi:10.1017/9781108608480 · Cited by 375
Comparison of text preprocessing methods
Chai, C. P. (2022) · Natural Language Engineering · doi:10.1017/s1351324922000213 · Cited by 303
“Researchers have attempted to mitigate these problems by borrowing information from a parallel corpus of another language, commonly in multilingual corpora for translation (Luo, Tinsley, and Lepage 2013; Zhang and Komachi 2018).”
BackgroundAn Efficient Method for Generating Synthetic Data for Low-Resource Machine Translation
Ngo, T.-V., Nguyen, P.-T., Nguyen, V. V., Ha, T.-L., & Nguyen, L.-M. (2022) · Applied Artificial Intelligence · doi:10.1080/08839514.2022.2101755 · Cited by 30
“As an otherwise idea, Zhang et al. (2018) use a forward model to predict the translations in the target language – called Self-learning.”
BackgroundMethodInfluentialWhy don’t people use character-level machine translation?
Libovický, J., Schmid, H., & Fraser, A. (2022) · Findings of the Association for Computational Linguistics: ACL (Findings of ACL) · doi:10.18653/v1/2022.findings-acl.194 · Cited by 35
“Zhang and Komachi (2018) argued that signs in logographic languages carry too much information and were able to improve the translation quality by segmenting Chinese and Japanese into sub-character units while keeping subword segmentation on the English side.”
BackgroundNguyen, M., Ngo, G. H., & Chen, N. F. (2019) · IEEE/ACM Transactions on Audio Speech and Language Processing · doi:10.1109/taslp.2019.2955246 · Cited by 23
“blem of out-of-vocabulary words and characters. However, alphabetic languages and languages with logographic origin are often treated the same way, disregarding their intrinsically marked differences [52]. Modeling logograph subunits can alleviate these issues since there are fewer subunits and they can be used to con…”
BackgroundInfluentialUsing Sub-character Level Information for Neural Machine Translation of Logographic Languages
Zhang, L., & Komachi, M. (2021) · ACM Transactions on Asian and Low-Resource Language Information Processing · doi:10.1145/3431727 · Cited by 11
“1 Character Decomposition and Composition Previous work adopted various resources to decompose characters (“Wubi” (five-stroke) input method,(1) [24], CNS11643 charset [32],(2) etc.”
BackgroundMethodSelf-citationGrammatically Derived Factual Relation Augmented Neural Machine Translation
Li, F., Zhu, J., Yan, H., & Zhang, Z. (2022) · Applied Sciences · doi:10.3390/app12136518 · Cited by 10
Wang, Z., Liu, X., & Zhang, M. (2022) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2022.emnlp-main.434 · Cited by 10
Chinese-Japanese Unsupervised Neural Machine Translation Using Sub-character Level InformationPreprint
Zhang, L., & Komachi, M. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1903.00149 · Cited by 10
“0001) of UNMT and supervised NMT systems (Zhang and Komachi, 2018).”
Self-citationChen, C. L., Dong, Y., Zambrano, C. C., Bencheqroun, H., Barwise, A. K., Hoffman, A., Nalaie, K., Qiu, Y., Boulekbache, O., & Niven, A. S. (2025) · BMC Medical Education · doi:10.1186/s12909-025-07452-9 · Cited by 8
“…for certain languages and differences in language structure between alphabetic (that is, languages in which letters are used construct words such as in Spanish or Ukrainian) and logographic languages (in which symbols or characters represent a unit of meaning, such as Mandarin) [18, 24, 25].”
Character Decomposition for Japanese-Chinese Character-Level Neural Machine Translation
Zhang, J., & Matsumoto, T. (2019) · International Conference on Asian Language Processing · doi:10.1109/ialp48816.2019.9037677 · Cited by 6
“For Chinese-Japanese NMT, the sub-character level information improved the translation performance [13], by using sub-character sequences on either the source or target side.”
MethodKoehn, P. (2020) · Cambridge University Press eBooks · doi:10.1017/9781108608480.003 · Cited by 5
Kim, H., Hirasawa, T., & Komachi, M. (2020) · Annual Meeting of the Association for Computational Linguistics: Student Research Workshop (ACL SRW) · doi:10.18653/v1/2020.acl-srw.11 · Cited by 3
“Zhang and Komachi (2018) demonstrated that higher translation accuracy can be obtained by decomposing Kanji into ideographic characters and strokes in Japanese-Chinese NMT. Stratos (2017) proposed a speech-parsing model for South Korean with character-level tokenization and decomposition into…”
BackgroundSelf-citationKoehn, P. (2020) · Cambridge University Press eBooks · doi:10.1017/9781108608480.009 · Cited by 4
Advancements in Neural Machine Translation Research Based on Language Knowledge
张, 津. (2023) · Artificial Intelligence and Robotics Research · doi:10.12677/airr.2023.122013 · Cited by 3
Nakano, Y., Saeki, T., Takamichi, S., Sudoh, K., & Saruwatari, H. (2023) · IEEE Spoken Language Technology Workshop (SLT) · doi:10.1109/slt54892.2023.10022739 · Cited by 3
“Starting with phrases [19] and words [20] and moving through subwords [21], characters [22], and sub-characters [23], some research proposes the use of pixels (i.”
BackgroundCharacter-Level Encoding based Neural Machine Translation for Hindi language
Rathod, D., Yadav, A. K., Kumar, M., & Yadav, D. (2025) · Neural Processing Letters · doi:10.1007/s11063-025-11718-0 · Cited by 2
“[45].”
Skansi, S. (2018) · Undergraduate topics in computer science · doi:10.1007/978-3-319-73004-2_9 · Cited by 2
Orthographic Codes and the Neighborhood Effect: Lessons from Information Theory
Tulkens, S., Sandra, D., & Daelemans, W. M. P. (2020) · Language Resources and Evaluation · Cited by 2
Inference-only sub-character decomposition improves translation of unseen logographic characters
Saunders, D., Feely, W., & Byrne, B. (2020) · Workshop on Asian Translation (WAT) · doi:10.18653/v1/2020.wat-1.21 · Cited by 2
“The baseline source and all target BPE vocabularies consist of character sequences, while the sub-character BPE vocabularies consist of sub-character sequences, following Zhang and Komachi (2018). For the lowerresource domains the unseen sets are held-out sentences containing logographic characters not in the baseline…”
InfluentialKoehn, P. (2020) · Cambridge University Press eBooks · doi:10.1017/9781108608480.008 · Cited by 2
Ngo, T.-V., Ha, T.-L., Nguyen, P.-T., & Nguyen, L.-M. (2019) · arXiv (Cornell University) · doi:10.5281/zenodo.3525489 · Cited by 2
Kondo, T., Donkers, J., Nishigori, H., Rovers, S., & Heeneman, S. (2026) · JMIR Medical Education · doi:10.2196/90064 · Cited by 1
“Although it varies by domain, Japanese is reported to have approximately 1.7 characters per word [47].”
InfluentialImproving English-Chinese translation using the Kolmogorov-Arnold Transformer
Nie, Y. (2025) · PeerJ Computer Science · doi:10.7717/peerj-cs.3139 · Cited by 1
A Non-image-based Subcharacter-level Method to Encode the Shape of Chinese Characters
Ke, Y., & Hagiwara, M. (2020) · Transactions of the Japanese Society for Artificial Intelligence · doi:10.1527/tjsai.c-j74 · Cited by 1
TMU Japanese-Chinese Unsupervised NMT System for WAT 2018 Translation Task.
Zhang, L., Zhao, Y., & Komachi, M. (2018) · Institutional Repositories DataBase (IRDB) · Cited by 1
Self-citationKoehn, P. (2020) · Cambridge University Press eBooks · doi:10.1017/9781108608480.018 · Cited by 3
Koehn, P. (2020) · Cambridge University Press eBooks · doi:10.1017/9781108608480.002 · Cited by 2
Koehn, P. (2020) · Cambridge University Press eBooks · doi:10.1017/9781108608480.004 · Cited by 1
Koehn, P. (2020) · Cambridge University Press eBooks · doi:10.1017/9781108608480.017 · Cited by 2
Compact and Robust Models for Japanese-English Character-level Machine Translation
Dai, J., & Yamaguchi, K. (2019) · Workshop on Asian Translation (WAT) · doi:10.18653/v1/d19-5202
“Zhang and Komachi (2018) proposed a sub-character level translation for Japanese and Chinese in which Kanji in Japanese and characters in Chinese are decomposed into ideographs or strokes.”
BackgroundKorean Neural Machine Translation Using Hierarchical Word Structure
Park, J., & Zhao, H. (2020) · International Conference on Asian Language Processing · doi:10.1109/ialp51396.2020.9310510
Koehn, P. (2020) · Cambridge University Press eBooks · doi:10.1017/9781108608480.015
Koehn, P. (2020) · Cambridge University Press eBooks · doi:10.1017/9781108608480.016
Koehn, P. (2020) · Cambridge University Press eBooks · doi:10.1017/9781108608480.013
Koehn, P. (2020) · Cambridge University Press eBooks · doi:10.1017/9781108608480.001
Koehn, P. (2020) · Cambridge University Press eBooks · doi:10.1017/9781108608480.007
Koehn, P. (2020) · Cambridge University Press eBooks · doi:10.1017/9781108608480.010
Koehn, P. (2020) · Cambridge University Press eBooks · doi:10.1017/9781108608480.005 · Cited by 2
Koehn, P. (2020) · Cambridge University Press eBooks · doi:10.1017/9781108608480.020
Koehn, P. (2020) · Cambridge University Press eBooks · doi:10.1017/9781108608480.011
Koehn, P. (2020) · Cambridge University Press eBooks · doi:10.1017/9781108608480.012
Koehn, P. (2020) · Cambridge University Press eBooks · doi:10.1017/9781108608480.006
Koehn, P. (2020) · Cambridge University Press eBooks · doi:10.1017/9781108608480.014
(2020) · Cambridge University Press eBooks · doi:10.1017/9781108608480.019
Koehn, P. (2020) · Cambridge University Press eBooks · doi:10.1017/9781108608480.021
Bidirectional LSTMs with Byte Pair Encoding in NMT for CLIR using English and Telugu Parallel Corpus
Raju, E. A. B. N. V. N. (2023) · International Journal on Recent and Innovation Trends in Computing and Communication · doi:10.17762/ijritcc.v11i9.8832
RETRACTED: Improving neural machine translation by word translations
Li, F., Chi, C., Yan, H., Zhang, Z., & Zhao, Z. (2024) · Journal of Intelligent & Fuzzy Systems · doi:10.3233/jifs-236170
“To allievate these issues, many researchers have made efforts to improve the translation quality by integrating prior knowledge, such as character or word structure [14, 16, 21, 22], phrase or chunk structure [17, 23] and syntactic structure [18, 24–27].”
RetractedBiLSTMs and BPE for English to Telugu CLIR
Raju, B. N. V. N. (2024) · Journal of Electrical Systems · doi:10.52783/jes.1798
“Sentences with a high frequency of words may result in OOV issues if resources are limited [16]–[18].”
BackgroundRaju, B. N. V. N., Satyanarayana, K. V. V. V., & Bhadri, R. M. S. V. S. (2024) · International Journal of Electronics and Communication Engineering · doi:10.14445/23488549/ijece-v11i5p117
“To address the issues caused by OOV [17, 18], the BPE mechanism is used.”
Improving Low-Resource Neural Machine Translation with Dependency Distance-Based Self-Attention
Yan, H., Li, F., Chen, Y., Chi, C., & Xie, P. (2025) · Lecture notes in computer science · doi:10.1007/978-981-95-0014-7_18
Incorporating Word Translations Into Neural Machine Translation
Li, F., Xie, P., Yan, H., Chi, C., Li, X., & Chen, W. (2025) · IEEE International Conference on High Performance Computing and Communications; IEEE International Conference on Smart City; IEEE International Conference on Data Science and Systems (HPCC/SmartCity/DSS) · doi:10.1109/hpcc67675.2025.00053
“Many researchers have made efforts to improve the translation quality of sequence-to-sequence models by integrating prior knowledge, such as character or word structure [10; 11; 12], phrase or chunk structure [13; 14] and syntactic structure [15; 16; 17; 18].”
Li, F., Ma, H., Yan, H., Chi, C., Xie, P., & Liu, B. (2025) · IEEE International Conference on High Performance Computing and Communications; IEEE International Conference on Smart City; IEEE International Conference on Data Science and Systems (HPCC/SmartCity/DSS) · doi:10.1109/hpcc67675.2025.00054
Metadata
- Authors
- Zhang, L., & Komachi, M.
- Affiliations
- Tokyo Metropolitan University (JP)
- Venue
- Proceedings of the Third Conference on Machine Translation: Research Papers
- Published
- 2018
- Pages
- 17-25
- OpenAlex
- W2963223057
- Cited by
- 53
- References
- 21
- Citations by year
- 2025: 6 · 2024: 3 · 2023: 3 · 2022: 6 · 2021: 1 · 2020: 27 · 2019: 5 · 2018: 2
- FWCI
- 3.57
- Type
- conference paper
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- cc-by
- Locations
- Conference on Machine Translation (WMT) (publishedVersion, cc-by)
- Funders
- Japan Society for the Promotion of Science (16K16117)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Natural Language Processing Techniques
- Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
- Physical Sciences › Computer Science › Artificial Intelligence › Speech Recognition and Synthesis
- Keywords
- neural machine translation, logographic languages, character decomposition, Chinese-English translation, Chinese-Japanese translation, subword units
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
