신경망기계번역 기술 진화와 번역품질 분석
A Study on the Evolution of Neural Machine Translation Technology and Translation Quality
0
조회
0
다운로드
11
인용
0
참고문헌
키워드
초록
There was a big technical progress in the research field of machine translation: the main approach has switched from statistical machine translation (SMT) to neural machine translation (NMT), leading to dramatic improvements in translation quality. Recently, another progress has been taking place from recurrent neural network(RNN)-based NMT to transformer-based NMT (T-NMT). As the performance of NMT has evolved, a lot of research papers for machine translation have been published in the field of interpretation and translation. Their main focus is on whether machine translation can replace human translation, and analyzing the quality of translation results. In this paper, we briefly explain the history of the machine translation research and review the mechanism of NMT. NMT is basically composed of three parts: encoder, attention mechanism, and decoder. Further we discuss the new transformer structure based on the encoder-decoder model. We also discuss the challenges in NMT and explain the research direction or solutions to the problems. Particular attention is given to the mistranslation of NMT, quality of the translation, and robustness against the noises in the training dataset as well as in the testing sentences. In order to test the performance of transformer-based NMT, we used the Google NMT (GNMT) service for 4 languages – Korean, English, German, and Japanese. We confirmed the robustness against sentences with noises. However, we found unexpected volatility of NMT models where the input sentence is semantically and syntactically correct, resulting in critical degradation of translation quality.
Abstract
There was a big technical progress in the research field of machine translation: the main approach has switched from statistical machine translation (SMT) to neural machine translation (NMT), leading to dramatic improvements in translation quality. Recently, another progress has been taking place from recurrent neural network(RNN)-based NMT to transformer-based NMT (T-NMT). As the performance of NMT has evolved, a lot of research papers for machine translation have been published in the field of interpretation and translation. Their main focus is on whether machine translation can replace human translation, and analyzing the quality of translation results. In this paper, we briefly explain the history of the machine translation research and review the mechanism of NMT. NMT is basically composed of three parts: encoder, attention mechanism, and decoder. Further we discuss the new transformer structure based on the encoder-decoder model. We also discuss the challenges in NMT and explain the research direction or solutions to the problems. Particular attention is given to the mistranslation of NMT, quality of the translation, and robustness against the noises in the training dataset as well as in the testing sentences. In order to test the performance of transformer-based NMT, we used the Google NMT (GNMT) service for 4 languages – Korean, English, German, and Japanese. We confirmed the robustness against sentences with noises. However, we found unexpected volatility of NMT models where the input sentence is semantically and syntactically correct, resulting in critical degradation of translation quality.
참고문헌 (34)
- 김순미, "AI시대 인간번역과 기계(NMT)번역의 공존-경영학 ‘확장(Augmentation)전략’ 중심", 통역과 번역, 20(2), pp.1~32 (2018), 한국통역번역학회[DOI]
- 김순미, "번역학계와 언어서비스업체(LSP)간 산학협력연구: ‘포스트에디팅 생산성’과 ‘기계번역 엔진 성능 비교’", 번역학연구, 20(1), pp.41~76 (2019), 한국번역학회[DOI]
- 마승혜, "기계화되기 어려운 인간 능력과 문학번역 능력 비교․고찰 및 논의", 통번역학연구, 21(3), pp.55~77 (2017), 통번역연구소[DOI]
- 마승혜, "문학작품 기계번역의 한계에 대한 상세 고찰", 통번역학연구, 22(3), pp.65~88 (2018), 통번역연구소[DOI]
- 윤호숙, "일한기계번역의 오류유형에 관한 고찰-일본소설을 중심으로-", 일어일문학연구, 107(1), pp.3~25 (2018), 한국일어일문학회[DOI]
- 이승일, "번역의 단계별 속성 및 번역주체에 대한 논의", 번역학연구, 18(5), pp.173~198 (2017), 한국번역학회[DOI]
- 이주리애, "인공신경망 기계 번역의 한일", 일한 번역 품질에 대한 예비연구-품질향상 검토와 교열코드 적용-, 2018, pp.20(1), 43~71[DOI]
- 이준호, "문학번역 적용을 위한 기계번역의 현주소", 통번역학연구, 23(1), pp.143~167 (2019), 통번역연구소[DOI]
- 이준호, "신경망기계번역의 객관적 평가를 위한 예비연구: 자동평가와 수동평가의 균형점", 통번역학연구, 23(3), pp.171~202 (2019), 통번역연구소[DOI]
- 지인영, "심층학습을 이용한 기계번역 기술과 정확도 연구", 인문언어, 19(2), pp.24~46 (2017), 국제언어인문학회
- 지인영, "비지도학습을 이용하는 다언어 기계번역시스템 기술분석", 정보산업공학논문집, 34, pp.77~88 (2019)
- 지인영, "사전학습 언어모델의 기술 분석 연구", 인문언어, 22(1), pp.111~133 (2020), 국제언어인문학회[DOI]
- 최희열, "신경망 기반 기계번역 모델의 이해", 정보과학회지, 37(2), pp.16~24 (2019)
- Bawden, R., "Evaluating Discourse Phenomena in Neural Machine Translation", Proceedings of NAACL-HLT, pp.1304~1313 (2018)
- Cheng, Y., "Robust Neural Machine Translation with Doubly Adversarial Inputs", arXiv:1906. 02443v1, ~ (2019)
- Cheng, Y., "Towards Robust Neural Machine Translation", arXiv:1805.06130v1, ~ (2018)
- Chu, C., "A Survey of Domain Adaptation for Neural Machine Translation", arXiv:1806.00258v1, ~ (2018)
- Clinchant, S., "On the use of BERT for Neural Machine Translation", arXiv:1909.12744v1, ~ (2019)
- Devlin, J., "BERT:Pre-training of Deep Bidirectional Transformers for Language Understanding", arXiv:1810.04805v2, ~ (2019)
- Fadaee, M., "The Unreasonable Volatility of Neural Machine Translation Models", arXiv:2005.12398v1, ~ (2020)
- Guoa, M., "Effective Parallel Corpus Mining using Bilingual Sentence Embeddings", arXiv:1807.11906v2, ~ (2018)
- Johnson, M., "Google’s Multilingual Neural Machine Translation System : Enabling Zero-Shot Translation", arXiv:1611.04558v2, ~ (2017)
- Khayrallah, H., "On the Impact of Various Types of Noise on Neural Machine Translation", Proceedings of the Workshop on Neural Machine Translation(WNMT) (2018)
- Koehn, P., "Six Challenges for Neural Machine Translation", arXiv:1706.03872, ~ (2017)
- Mikolov, T., "Distributed representations of words and phrases and their compositionality", Advances in Neural Information Processing Systems, 26, pp.3111~3119 (2013)
- Raffel, C., "Exploring the limits of transfer learning with a unified text-to-text transformer", arXiv:1910.10683v3, ~ (2019)
- Schwenk, H., "CCMatrix: Mining Billions of High-Quality Parallel Sentences on the WEB", arXiv:1911.04944v2, ~ (2020)
- Sennrich, R., "Improving Neural Machine Translation Models with Monolingual Data", arXiv:1511.06709, ~ (2016)
- Song, K., "Mass : Masked sequence to sequence pre-training for language generation", arXiv:1905.02450v5, ~ (2019)
- Susanto, R. H., "Lexically Constrained Neural Machine Translation with Levenshtein Transformer", arXiv:2004.12681v1, ~ (2020)
- Vaswani, A., "Attention Is All You Need", NIPS 2017 (2017)
- Wang, R., "A Survey of Advances and Challenges in Unsupervised Neural Machine Translation", 言語処理学会第26回年次大会(NLP2020) (2020)
- Wang, W., "Denoising Neural Machine Translation Training with Trusted Data and Online Data Selection", arXiv:1809.00068v1, ~ (2018)
- Zhang, J., "Neural Machine Translation: Challenges, Progress and Future", arXiv:2004.05809v1, ~ (2020)