Abstract
This paper demonstrates our machine translation systems for the CCMT 2020, which is composed of four parts. The last three parts report our results in the contest, each respectively focuses on English-Chinese bi-direction translation, Japanese-Chinese-English multi-lingual translation (patent domain), and Chinese minority languages to Mandarin Chinese translation. In each part, we will demonstrate our work on data pre-processing, model training as well as the application of general techniques, such as back-translation, ensemble and reranking. Besides, during our experiments, we surprisingly found that simply applying different Chinese word segmentation tools on low-resource corpora could bring obvious benefit across different tasks, and we will separate an independent section to discuss this finding. Among the 7 directions we participated in, we ranked the first in 6 tasks (For the corpus filtering task, we ranked first in the 500 million words sub-task) and the second for the rest.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Similar content being viewed by others
Notes
- 1.
- 2.
- 3.
Including datasets released by WMT 2020, which are allowed in CCMT 2020.
- 4.
- 5.
- 6.
- 7.
- 8.
More accurately, morpheme-based model.
- 9.
For Uighur \(\rightarrow \) Mandarin task we didn’t filter the corpus according to alignment information, since we find sometimes a Mandarin word can be a long phrase in Uighur. e.g. “
” (rule of law) is officially translated to “qanun arqiliq idare qilish”. (Uighur here is transliterated by Uighur Latin alphabet (ULY)). For Tibetan, alignment information is calculated on a character-level corpus, means not only the Tibetan data is segmented by morpheme, but also the Mandarin data is split into characters.
” (rule of law) is officially translated to “qanun arqiliq idare qilish”. (Uighur here is transliterated by Uighur Latin alphabet (ULY)). For Tibetan, alignment information is calculated on a character-level corpus, means not only the Tibetan data is segmented by morpheme, but also the Mandarin data is split into characters.References
Vaswani, A., et al.: Attention is all you need. In: Advances in Neural Information Processing Systems (NeurIPS 2017), pp. 5998–6008 (2017)
Xue, Z., Zhang, Q., Li, X., Dang, D., Zhang, G., Hao, J.: OPPO machine translation system. In: Proceedings of the 15th China Conference on Machine Translation (CCMT 2019) (2019). (in Chinese)
Zhang, Q., et al.: OPPO’s machine translation system for the IWSLT 2020 open domain translation task. In: Proceedings of the 17th International Conference on Spoken Language Translation (IWSLT 2020), pp. 114–121, July 2020
Kingma, D.P., Ba, J.: Adam: a method for stochastic optimization. arXiv preprint arXiv:1412.6980 (2014)
Ott, M., et al.: fairseq: a fast, extensible toolkit for sequence modeling. In: Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics (Demonstrations), pp. 48–53, June 2019
Sennrich, R., Haddow, B., Birch, A.: Improving neural machine translation models with monolingual data. In: Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) (ACL 2016), pp. 86–96, August 2016
Hoang, V.C.D., Koehn, P., Haffari, G., Cohn, T.: Iterative back-translation for neural machine translation. In: Proceedings of the 2nd Workshop on Neural Machine Translation and Generation (NMT 2018), pp. 18–24, July 2018
Edunov, S., Ott, M., Auli, M., Grangier, D.: Understanding back-translation at scale. In: Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing (EMNLP 2018), pp. 489–500 (2018)
Freitag, M., Al-Onaizan, Y., Sankaran, B.: Ensemble distillation for neural machine translation. arXiv preprint arXiv:1702.01802 (2017)
Heafield, K., Pouzyrevsky, I., Clark, J.H., Koehn, P.: Scalable modified Kneser-Ney language model estimation. In: Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics (ACL 2013) (Volume 2: Short Papers), pp. 690–696, August 2013
Cherry, C., Foster, G.: Batch tuning strategies for statistical machine translation. In: Proceedings of the 2012 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT 2012), pp. 427–436, June 2012
Li, X., Meng, Y., Sun, X., Han, Q., Yuan, A., Li, J.: Is word segmentation necessary for deep learning of Chinese representations? In: Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics (ACL 2019), pp. 3242–3252, July 2019
Luo, R., Xu, J., Zhang, Y., Ren, X., Sun, X.: PKUSEG: a toolkit for multi-domain Chinese word segmentation. arXiv preprint arXiv:1906.11455 (2019)
Junczys-Dowmunt, M., et al.: Marian: fast neural machine translation in C++. In: Proceedings of ACL 2018, System Demonstrations, pp. 116–121, July 2018
Hu, B., Han, A., Zhang, Z., Huang, S., Ju, Q.: Tencent minority-mandarin translation system. In: Huang, S., Knight, K. (eds.) CCMT 2019. CCIS, vol. 1104, pp. 93–104. Springer, Singapore (2019). https://doi.org/10.1007/978-981-15-1721-1_10
Johnson, M., et al.: Google’s multilingual neural machine translation system: enabling zero-shot translation. Trans. Assoc. Comput. Linguist. (TACL) 5, 339–351 (2017)
Dyer, C., Chahuneau, V., Smith, N.A.: A simple, fast, and effective reparameterization of IBM model 2. In: Proceedings of the 2013 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT 2013), pp. 644–648, June 2013
Yee, K., Dauphin, Y., Auli, M.: Simple and effective noisy channel modeling for neural machine translation. In: Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP 2019), pp. 5700–5705, November 2019
Biçici, E., Yuret, D.: Instance selection for machine translation using feature decay algorithms. In: Proceedings of the Sixth Workshop on Statistical Machine Translation (WMT 2011), pp. 272–283, July 2011
Post, M.: A call for clarity in reporting BLEU scores. In: Proceedings of the Third Conference on Machine Translation: Research Papers (WMT 2018), pp. 186–191, October 2018
Huang, C., Zhao, H.: Chinese word segmentation: a decade review. J. Chin. Inf. Process. 21(3), 8–19 (2007)
Zhao, H., Cai, D., Huang, C., Kit, C.: Chinese word segmentation: another decade review (2007–2017). arXiv preprint arXiv:1901.06079 (2019)
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2020 Springer Nature Singapore Pte Ltd.
About this paper
Cite this paper
Shi, T., Zhang, Q., Wang, X., Li, X., Xue, Z., Hao, J. (2020). Description and Findings of OPPO’s Machine Translation Systems for CCMT 2020. In: Li, J., Way, A. (eds) Machine Translation. CCMT 2020. Communications in Computer and Information Science, vol 1328. Springer, Singapore. https://doi.org/10.1007/978-981-33-6162-1_8
Download citation
DOI: https://doi.org/10.1007/978-981-33-6162-1_8
Published:
Publisher Name: Springer, Singapore
Print ISBN: 978-981-33-6161-4
Online ISBN: 978-981-33-6162-1
eBook Packages: Computer ScienceComputer Science (R0)