Data-driven Machine Translation Research Articles

Integrating professional machine translation literacy and data literacy

AbstractThe data-driven paradigm of neural machine translation is a powerful translation technology based on state-of-the art approaches in artificial intelligence research. This technology is employed extensively in the professional translation process, requiring an adequate degree of machine translation literacy on the part of professional translators. At the same time, the increasing datafication to be observed in modern society – both in private as well as in professional contexts – contributes to the rise in prominence of another digital literacy, i. e., data literacy, which is also of high relevance with regard to data-driven machine translation. The present paper analyses and discusses in detail the individual dimensions and subdimensions of professional machine translation literacy and data literacy and attempts to integrate the two concepts. It thereby lays the theoretical foundation for a didactic project concerned with teaching data literacy in its machine translation-specific form to students in the fields of translation and specialised communication studies.

Research on Traditional Mongolian-Chinese Neural Machine Translation Based on Dependency Syntactic Information and Transformer Model

Neural machine translation (NMT) is a data-driven machine translation approach that has proven its superiority in large corpora, but it still has much room for improvement when the corpus resources are not abundant. This work aims to improve the translation quality of Traditional Mongolian-Chinese (MN-CH). First, the baseline model is constructed based on the Transformer model, and then two different syntax-assisted learning units are added to the encoder and decoder. Finally, the encoder’s ability to learn Traditional Mongolian syntax is implicitly strengthened, and the knowledge of Chinese-dependent syntax is taken as prior knowledge to explicitly guide the decoder to learn Chinese syntax. The average BLEU values measured under two experimental conditions showed that the proposed improved model improved by 6.706 (45.141–38.435) and 5.409 (41.930–36.521) compared with the baseline model. The analysis of the experimental results also revealed that the proposed improved model was still deficient in learning Chinese syntax, and then the Primer-EZ method was introduced to ameliorate this problem, leading to faster convergence and better translation quality. The final improved model had an average BLEU value increase of 9.113 (45.634–36.521) compared with the baseline model at experimental conditions of N = 5 and epochs = 35. The experiments showed that both the proposed model architecture and prior knowledge could effectively lead to an increase in BLEU value, and the addition of syntactic-assisted learning units not only corrected the initial association but also alleviated the long-term dependence between words.

A Review and evaluation of Machine Translation methods for Lumasaaba

Natural Language Processing for under-resourced languages is now a mainstream research area. However, there are limited studies on Natural Language Processing applications for many indigenous East African languages. As a contribution to covering the current gap of knowledge, this paper focuses on evaluating the application of well-established machine translation methods for one heavily under-resourced indigenous East African language called Lumasaaba. Specifically, we review the most common machine translation methods in the context of Lumasaaba including both rule-based and data-driven methods. Then we apply a state of the art data-driven machine translation method to learn models for automating translation between Lumasaaba and English using a very limited data set of parallel sentences. Automatic evaluation results show that a transformer-based Neural Machine Translation model architecture leads to consistently better BLEU scores than the recurrent neural network-based models. Moreover, the automatically generated translations can be comprehended to a reasonable extent and are usually associated with the source language input.

Low-Resource Multi-Domain Machine Translation for Spanish-Farsi: Neural or Statistical?

An important prerequisite for data-driven Machine Translation (MT) systems is the availability of high-quality training data. Corpus-based MT systems requiring domain specificity additionally require the selection of a large training dataset and application of proper domain adaptation techniques. This paper demonstrates the portability of MT systems to multiple domains under low-resource conditions for the target language, where target-language domain data are not available. We refer to this case as a bilingually low-resource language pair. We compare the performance of an attention-based Neural MT (NMT) system to that of a phrase-based Statistical MT (SMT) system under this condition, training each on parallel corpora consisting of different domains. Experimental results on Spanish-Farsi as a bilingually low-resource language pair show that the SMT paradigm still outperforms that of NMT.

Spanish-Turkish Low-Resource Machine Translation: Unsupervised Learning vs Round-Tripping

The quality of data-driven Machine Translation (MT) strongly depends on the quantity as well as the quality of the training dataset. However, collecting a large set of training parallel texts is not easy in practice. Although various approaches have already been proposed to overcome this issue, the lack of large parallel corpora still poses a major practical problem for many language pairs. Since monolingual data plays an important role in boosting fluency for Neural MT (NMT) models, this paper investigates and compares the performance of two learning-based translation approaches for Spanish-Turkish translation as a low-resource setting in case we only have access to large sets of monolingual data in each language; 1) Unsupervised Learning approach, and 2) Round-Tripping approach. Either approach completely removes the need for bilingual data or enables us to train the NMT system relying on monolingual data only. We utilize an Attention-based NMT (Attentional NMT) model, which leverages a careful initialization of the parameters, the denoising effect of language models, and the automatic generation of bilingual data. Our experimental results demonstrate that the Unsupervised Learning approach outperforms the Round-Tripping approach in Spanish-Turkish translation and vice versa. These results confirm that the Unsupervised Learning approach is still a reliable learning-based translation technique for Spanish-Turkish low-resource NMT.

Efficient Use of Resources for Statistical Machine Translation

<div class="page" title="Page 1"><div class="layoutArea"><div class="column"><p><span>Machine translation has great potential to expand the audience for ever increasing digital collections. Success of data driven machine translation systems is governed by the volume of parallel data on which these systems are being modelled. The languages which do not have such resources in huge quantity, the optimum utilisation of them can only be assured through their quality. Morphologically rich language like Hindi poses further challenge, due to </span><span>having more number of orthographic inflections for a given word and presence of non-standard word spellings in </span><span>the corpus. This increases the chances of getting more number of words which are unseen in the training corpus. In this paper, the objective is to reduce redundancy of available corpus and utilise the other resources as well, to make best use of resources. Reduction in number of words unseen to the translation model is achieved through text noise removal, spell normalisation and utilising English WordNet (EWN). The test case presented here is for English-Hindi language pair. The results achieved are promising and set example for other morphological rich languages to optimise the resources to improve the performance of the translation system. </span></p></div></div></div>

Efficient Use of Resources for Statistical Machine Translation

<div class="page" title="Page 1"><div class="layoutArea"><div class="column"><p><span>Machine translation has great potential to expand the audience for ever increasing digital collections. Success of data driven machine translation systems is governed by the volume of parallel data on which these systems are being modelled. The languages which do not have such resources in huge quantity, the optimum utilisation of them can only be assured through their quality. Morphologically rich language like Hindi poses further challenge, due to </span><span>having more number of orthographic inflections for a given word and presence of non-standard word spellings in </span><span>the corpus. This increases the chances of getting more number of words which are unseen in the training corpus. In this paper, the objective is to reduce redundancy of available corpus and utilise the other resources as well, to make best use of resources. Reduction in number of words unseen to the translation model is achieved through text noise removal, spell normalisation and utilising English WordNet (EWN). The test case presented here is for English-Hindi language pair. The results achieved are promising and set example for other morphological rich languages to optimise the resources to improve the performance of the translation system. </span></p></div></div></div>

Manual labour: tackling machine translation for sign languages

This article explores the application of data-driven machine translation (MT) to sign languages (SLs). The provision of an SL MT system can facilitate communication between Deaf and hearing people by translating information into the native and preferred language of the individual. In this paper we address data-driven SL MT predominantly for Irish SL (ISL) but also for German SL (DGS/Deutsche Gebardensprache). We take two different purpose-built corpora to feed our MaTrEx MT system and in a set of experiments translating both to and from the SLs, we investigate the effects of SL data on statistical MT (SMT). Exploiting the bidirectionality of the MaTrEx system, we demonstrate how additional modules, such as recognition and SL animation, can potentially build a full SL MT model for spoken and SL communication in addition to promising evaluation scores. A secondary focus of the article is on the two main issues affecting SL MT, those of transcription and evaluation. We offer a discussion on both these common problems before concluding.

Cunei: open-source machine translation with relevance-based models of each translation instance

The Cunei machine translation platform is an open-source system for data-driven machine translation. Our platform is a synthesis of the traditional example-based MT (EBMT) and statistical MT (SMT) paradigms. What makes Cunei unique is that it measures the relevance of each translation instance with a distance function. This distance function, represented as a log-linear model, operates over one translation instance at a time and enables us to score the translation instance relative to the specified input and/or the current target hypothesis. We describe how our system, Cunei, scores features individually for each translation instance and how it efficiently performs parameter tuning over the entire feature space. We also compare Cunei with three other open-source MT systems (Moses, CMU-EBMT, and Marclator). In our experiments involving Korean---English and Czech---English translation Cunei clearly outperforms the traditional EBMT and SMT systems.