Multi-source Neural Automatic Post-Editing: FBK’s participation in
            the WMT 2017 APE shared task

Chatterjee, Rajen; Farajian, M. Amin; Negri, Matteo; Turchi, Marco; Srivastava, Ankit; Pal, Santanu Kumar

doi:10.18653/v1/w17-4773

Cited by 32 publications

(25 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The translations can then be post-edited, a process that is less labor-intensive and cheaper compared to translating from scratch. Multi-source NMT has been used for post-editing where the translated sentence is used as an additional source, leading to improvements [18]. Multi-source NMT has also been used for system combination, which combines NMT and SMT outputs to improve translation performance [166].…”

Section: Multi-source Nmtmentioning

confidence: 99%

A Survey of Multilingual Neural Machine Translation

2020

View full text Add to dashboard Cite

We present a survey on multilingual neural machine translation (MNMT), which has gained a lot of traction in recent years. MNMT has been useful in improving translation quality as a result of translation knowledge transfer (transfer learning). MNMT is more promising and interesting than its statistical machine translation counterpart, because end-to-end modeling and distributed representations open new avenues for research on machine translation. Many approaches have been proposed to exploit multilingual parallel corpora for improving translation quality. However, the lack of a comprehensive survey makes it difficult to determine which approaches are promising and, hence, deserve further exploration. In this article, we present an indepth survey of existing literature on MNMT. We first categorize various approaches based on their central use-case and then further categorize them based on resource scenarios, underlying modeling principles, coreissues, and challenges. Wherever possible, we address the strengths and weaknesses of several techniques by comparing them with each other. We also discuss the future directions for MNMT. This article is aimed towards both beginners and experts in NMT. We hope this article will serve as a starting point as well as a source of new ideas for researchers and engineers interested in MNMT.

show abstract

Section: Multi-source Nmtmentioning

confidence: 99%

A Survey of Multilingual Neural Machine Translation

2020

View full text Add to dashboard Cite

show abstract

“…Our model outperforms the best performing system at the last round of the shared task (Chatterjee et al, 2017), with improvements up to -1.27 TER and +1.23 BLEU on the PBSMT development set. Although we are using more out-ofdomain data, it is interesting to note that these scores are obtained with a much simpler architecture, which does not require to ensemble n models and to train a re-ranker.…”

Section: Resultsmentioning

confidence: 88%

“…In the last few years, the APE shared tasks at WMT (Bojar et al, 2015(Bojar et al, , 2016(Bojar et al, , 2017 have renewed the interests in this topic and boosted the technology around it. Moving from the phrase-based approaches used in the first editions of the task , last year the multi-source neural models (Chatterjee et al, 2017;Junczys-Dowmunt and Grundkiewicz, 2017;Hokamp, 2017) have shown their capability to significantly improve the output of a PBSMT system. These APE systems shared several features and implementation choices, namely: 1) an RNN-based architecture, 2) the use of large artificial corpora for training, 3) model ensembling techniques, 4) parameter optimization based on Maximum Likelihood Estimation (MLE) and 5) vocabulary reduction using the Byte Pair Encoding (BPE) technique.…”

Section: Introductionmentioning

confidence: 99%

Multi-source transformer with combined losses for automatic post editing

Tebbifakhr¹,

Agrawal²,

Negri³

et al. 2018

Proceedings of the Third Conference on Machine Translation: Shared Task Papers

Self Cite

View full text Add to dashboard Cite

Recent approaches to the Automatic Postediting (APE) of Machine Translation (MT) have shown that best results are obtained by neural multi-source models that correct the raw MT output by also considering information from the corresponding source sentence. To this aim, we present for the first time a neural multi-source APE model based on the Transformer architecture. Moreover, we employ sequence-level loss functions in order to avoid exposure bias during training and to be consistent with the automatic evaluation metrics used for the task. These are the main features of our submissions to the WMT 2018 APE shared task, where we participated both in the PBSMT subtask (i.e. the correction of MT outputs from a phrase-based system) and in the NMT subtask (i.e. the correction of neural outputs). In the first subtask, our system improves over the baseline up to-5.3 TER and +8.23 BLEU points ranking second out of 11 submitted runs. In the second one, characterized by the higher quality of the initial translations, we report lower but statistically significant gains (up to-0.38 TER and +0.8 BLEU), ranking first out of 10 submissions.

show abstract

“…Considering the MT output as a source sentence and the post-edited output as a target sentence, this problem can be cast as a monolingual translation task and be addressed with different MT solutions (Simard et al, 2007;Pal et al, 2016). However, it has been proven that better performance can be obtained by not only using the raw output of the MT system but also by leveraging the source text (Chatterjee et al, 2017). In the last round of the APE shared task (Chatterjee et al, 2018a), the top-ranked systems (Tebbifakhr et al, 2018;Junczys-Dowmunt and Grundkiewicz, 2018) were based on Transformer (Vaswani et al, 2017), the state-of-the-art architecture in neural MT (NMT), with two encoders to encode both source text and MT output.…”

Section: Introductionmentioning

confidence: 99%

Effort-Aware Neural Automatic Post-Editing

Tebbifakhr¹,

Negri²,

Turchi³

2019

Proceedings of the Fourth Conference on Machine Translation (Volume 3: Shared Task Papers, Day 2)

Self Cite

View full text Add to dashboard Cite

For this round of the WMT 2019 APE shared task, our submission focuses on addressing the "over-correction" problem in APE. Overcorrection occurs when the APE system tends to rephrase an already correct MT output, and the resulting sentence is penalized by a reference-based evaluation against human post-edits. Our intuition is that this problem can be prevented by informing the system about the predicted quality of the MT output or, in other terms, the expected amount of needed corrections. For this purpose, following the common approach in multilingual NMT, we prepend a special token to the beginning of both the source text and the MT output indicating the required amount of post-editing. Following the best submissions to the WMT 2018 APE shared task, our backbone architecture is based on multi-source Transformer to encode both the MT output and the corresponding source text. We participated both in the English-German and English-Russian subtasks. In the first subtask, our best submission improved the original MT output quality up to +0.98 BLEU and-0.47 TER. In the second subtask, where the higher quality of the MT output increases the risk of over-correction, none of our submitted runs was able to improve the MT output.

show abstract

Multi-source Neural Automatic Post-Editing: FBK’s participation in the WMT 2017 APE shared task

Cited by 32 publications

References 19 publications

A Survey of Multilingual Neural Machine Translation

A Survey of Multilingual Neural Machine Translation

Multi-source transformer with combined losses for automatic post editing

Effort-Aware Neural Automatic Post-Editing

Contact Info

Product

Resources

About