Adapting Neural Machine Translation with Parallel Synthetic Data

By Mara Chinea-Rios, Álvaro Peris, Francisco Casacuberta. 2017

65 citations recorded by the source corpus.

Recent works have shown that the usage of a synthetic parallel corpus can be effectively exploited by a neural machine translation system. In this paper, we propose a new method for adapting a general neural machine translation system to a specific task, by exploiting synthetic data. The method consists in selecting, from a large monolingual pool of sentences in the source language, those instances that are more related to a given test set. Next, this selection is automatically translated and the general neural machine translation sys-tem is fine-tuned with these data. For evaluating the adaptation method, we first conducted experiments in two controlled domains, with common and well-studied corpora. Then, we evaluated our proposal on a real e-commerce task, yielding consistent improvements in terms of translation quality.

Subfields: Artificial Intelligence, Natural Language Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Tao Qin
2. Tie-Yan Liu
3. Lijun Wu
4. Yingce Xia
5. Xipeng Qiu
6. Xu Tan
7. Eduard H. Hovy
8. Lei Li

In Natural Language Processing

1. Tao Qin
2. Tie-Yan Liu
3. Lijun Wu
4. Yingce Xia
5. Xipeng Qiu
6. J. Lai
7. Eduard H. Hovy
8. Varun Prashant Gangal

Citations in the corpus

• A Survey of Data Augmentation Approaches for NLP (2021, methodology): “Chinea-Ríos et al. (2017) Propose DA method to create synthetic data by leveraging the embedding representation of sentences.”
• Exploiting Monolingual Data at Scale for Neural Machine Translation (2019, background): “…data is an important research topic for NMT and there are plenty of works studying this problem (Gulcehre et al., 2015, 2017; Sennrich et al., 2016a; He et al., 2016; Zhang and Zong, 2016; Zhang et al., 2018; Cheng et al., 2016; Wang et al., 2018; Chinea-Rios et al., 2017; Wu et al., 2018).”
• The Source-Target Domain Mismatch Problem in Machine Translation (2019, unknown): “Zhang and Zong (2016) for neural MT. In our work, we consider the iterative variant proposed by He et al. (2020), whereby all model parameters are subject to training and noise is added to the input. Chinea-Rios et al. (2017) showed that self-training can be used to adapt to a different domain after”
• Abstract Text Summarization: A Low Resource Challenge (2019, unknown): “The use of synthetic data along with the real data is one of the popular approaches followed in machine translation domain for the low resource conditions to improve the translation quality (Bojar and Tamchyna, 2011; Hoang et al., 2018; Chinea-Rıos et al., 2017).”
• The practical ethics of bias reduction in machine translation: why domain adaptation is better than data debiasing (2021, methodology): “A cruder, but simpler, approach involves creating counterfactual translations automatically using an existing MT system (Chinea-Rìos et al. 2017).”
• Finding Sparse Structures for Domain Specific Neural Machine Translation (2021, methodology): “Data-centric methods focus on selecting or generating target domain data from general domain corpora, which is effective and well explored (Axelrod, He, and Gao 2011; Chinea-R´ıos, Peris, and Casacuberta 2017; Zeng et al. 2019).”