This paper presents the results of the premier shared task organized alongside the Conference on Machine Translation (WMT) 2019. Participants were asked to build machine translation systems for any of 18 language pairs, to be evaluated on a test set of news stories. The main metric for this task is human judgment of translation quality. The task was also opened up to additional test suites to probe specific aspects of translation.

/pdf/findings-of-the-2019-conference-on-machine-translation-wmt19-4hs5zeff36.pdf

Findings of the 2019 Conference on Machine Translation (WMT19)

THIS year there have been completed two very notable works in the field of linguistic science: one is the “New English Dictionary,” finished after seventy years of labour; the other is the great “Linguistic Survey of India,” of which the last volume to be published is now before us. No scholar can work at any problem connected with the history of English without constant appeal to the Dictionary, and no scholar can work at any problem connected with the languages of India without constant appeal to the Survey. Indeed, we can scarcely recall during the last fifteen years an article or book on the history of any of these languages (and shortly it will be seen how numerous and diverse they are) in which reference has not been made to the facts set forth in the Survey, often for the first time. Linguistic Survey of India. By Sir George Abraham Grierson. Vol. 1, Part 1: Introductory. Pp. xviii + 517. (Calcutta: Government of India Central Publication Branch; London: High Commissioner for India, 1927.) 11.12 rupees; 19s.

Linguistic Survey of India

Large-scale temporal gene expression mapping of central nervous system development

Sarcasm detection in mash-up language using soft-attention based bi-directional LSTM and feature-rich CNN

This paper presents the results of the news translation task and the similar language translation task, both organised alongside the Conference on Machine Translation (WMT) 2020. In the news task, participants were asked to build machine translation systems for any of 11 language pairs, to be evaluated on test sets consisting mainly of news stories. The task was also opened up to additional test suites to probe specific aspects of translation. In the similar language translation task, participants built machine translation systems for translating between closely related pairs of languages.

Findings of the 2020 Conference on Machine Translation (WMT20)

Machine translation helps resolve language incomprehensibility issues and eases interaction among people from varying linguistic backgrounds. Although corpus-based approaches (statistical and neural) offer reasonable translation accuracy for large-sized corpus, robustness of such approaches lie in their ability to adapt to low-resource languages, which confront unavailability of large-sized corpus. In this paper, prediction aptness of two approaches has been meticulously explored in the context of Mizo, a low-resource Indian language. Translations predicted by the two approaches have been comparatively and adequately analyzed on a number of grounds to infer their strengths and weaknesses, particularly in low-resource scenarios.

English–Mizo Machine Translation using neural and statistical approaches

With the widespread use of Machine Trans-lation (MT) techniques, attempt to minimizecommunication gap among people from di-verse linguistic backgrounds. We have par-ticipated in Workshop on Asian Transla-tion 2019 (WAT2019) multi-modal translationtask. There are three types of submissiontrack namely, multi-modal translation, Hindi-only image captioning and text-only transla-tion for English to Hindi translation. The mainchallenge is to provide a precise MT output.The multi-modal concept incorporates textualand visual features in the translation task. Inthis work, multi-modal translation track re-lies on pre-trained convolutional neural net-works (CNN) with Visual Geometry Grouphaving 19 layered (VGG19) to extract imagefeatures and attention-based Neural MachineTranslation (NMT) system for translation.The merge-model of recurrent neural network(RNN) and CNN is used for the Hindi-onlyimage captioning. The text-only translationtrack is based on the transformer model of theNMT system. The official results evaluated atWAT2019 translation task, which shows thatour multi-modal NMT system achieved Bilin-gual Evaluation Understudy (BLEU) score20.37, Rank-based Intuitive Bilingual Eval-uation Score (RIBES) 0.642838, Adequacy-Fluency Metrics (AMFM) score 0.668260 forchallenge test data and BLEU score 40.55,RIBES 0.760080, AMFM score 0.770860 forevaluation test data in English to Hindi multi-modal translation respectively.

/pdf/english-to-hindi-multi-modal-neural-machine-translation-and-2m5bie51ik.pdf

English to Hindi Multi-modal Neural Machine Translation and Hindi Image Captioning

Neural machine translation (NMT) attracts attention to the machine translation (MT) community because of its potential to persist sequential data over variable lengths of input and output sentences. With the attention mechanism, the NMT system achieves state-of-the-art technique which allows focusing on specific input vectors of the input sentence instead of learning a single vector for each input sentence. Although, the NMT system improves translation quality by handling long-term dependency issue and facilitate the ability of contextual analysis, and it requires adequate parallel train corpus, which is difficult in the context of low resource languages. This paper contributes an Assamese–English parallel corpus and builds two NMT systems, sequence-to-sequence recurrent neural network (RNN) with attention mechanism (NMT-1) and a transformer model having self-attention mechanism (NMT-2). The NMT-2 system has achieved a higher bilingual evaluation understudy (BLEU) score of 10.03 for English-to-Assamese and BLEU score 13.10 for Assamese-to-English translation, respectively.

Neural Machine Translation for Low Resource Assamese–English

With the extensive use of Machine Translation (MT) technology, there is progressively interest in directly translating between pairs of similar languages. Because the main challenge is to overcome the limitation of available parallel data to produce a precise MT output. Current work relies on the Neural Machine Translation (NMT) with attention mechanism for the similar language translation of WMT19 shared task in the context of Hindi-Nepali pair. The NMT systems trained the Hindi-Nepali parallel corpus and tested, analyzed in Hindi ⇔ Nepali translation. The official result declared at WMT19 shared task, which shows that our NMT system obtained Bilingual Evaluation Understudy (BLEU) score 24.6 for primary configuration in Nepali to Hindi translation. Also, we have achieved BLEU score 53.7 (Hindi to Nepali) and 49.1 (Nepali to Hindi) in contrastive system type.

/pdf/neural-machine-translation-hindi-nepali-3v2g2atqkp.pdf

Neural Machine Translation: Hindi-Nepali

The work presented in this paper ascertains role of Long Sort-Term Memory (LSTM) neural network in Math Information Retrieval (MIR). Motivated from promising performances of the LSTM for sequence-to-sequence tasks, an LSTM based Formula Entailment (LFE) module is implemented for recognizing entailment between mathematical user query and document formulae. The LFE module is trained and validated using a symbol level Math Formula Entailment (MENTAIL) dataset. The relevance of a document is determined by the fraction of document formulae which entail the user query. A reasonable score of 0.45 for the P_5 evaluation measure substantiates competence of the implemented MIR system in retrieving relevant documents corresponding to a mathematical user query.

Partha Pakray

Papers

English–Mizo Machine Translation using neural and statistical approaches

English to Hindi Multi-modal Neural Machine Translation and Hindi Image Captioning

Neural Machine Translation for Low Resource Assamese–English

Neural Machine Translation: Hindi-Nepali

LSTM Neural Network Based Math Information Retrieval