A significant challenge for machine translation (MT) is the phenomena of dropped pronouns (DPs), where certain classes of pronouns are frequently dropped in the source language but should be retained in the target language. In response to this common problem, we propose a semi-supervised approach with a universal framework to recall missing pronouns in translation. Firstly, we build training data for DP generation in which the DPs are automatically labelled according to the alignment information from a parallel corpus. Secondly, we build a deep learning-based DP generator for input sentences in decoding when no corresponding references exist. More specifically, the generation has two phases: (1) DP position detection, which is modeled as a sequential labelling task with recurrent neural networks; and (2) DP prediction, which employs a multilayer perceptron with rich features. Finally, we integrate the above outputs into our statistical MT (SMT) system to recall missing pronouns by both extracting rules from the DP-labelled training data and translating the DPgenerated input sentences. To validate the robustness of our approach, we investigate our approach on both Chinese-English and Japanese-English corpora extracted from movie subtitles. Compared with an SMT baseline system, experimental results show that our approach achieves a significant improvement of +1.58 BLEU points in translation performance with 66% F-score for DP generation accuracy for Chinese-English, and nearly +1 BLEU point with 58% F-score for Japanese-English. We believe that this work could help both MT researchers and industries to boost the performance of MT systems between pro-drop and non-pro-drop languages.
Recent years have seen a surge of interest in dialogue translation, which is a significant application task for machine translation (MT) technology. However, this has so far not been extensively explored due to its inherent characteristics including data limitation, discourse properties and personality traits. In this article, we give the first comprehensive review of dialogue MT, including well-defined problems (e.g., 4 perspectives), collected resources (e.g., 5 language pairs and 4 sub-domains), representative approaches (e.g., architecture, discourse phenomena and personality) and useful applications (e.g., hotel-booking chat system). After systematical investigation, we also build a state-of-the-art dialogue NMT system by leveraging a breadth of established approaches such as novel architectures, popular pre-training and advanced techniques. Encouragingly, we push the state-of-the-art performance up to 62.7 BLEU points on a commonly-used benchmark by using mBART pre-training. We hope that this survey paper could significantly promote the research in dialogue MT.
Translating conversational text, in particular task-oriented dialogues, is an important application task for machine translation technology. However, it has so far not been extensively explored due to its inherent characteristics including data limitation, discourse, informality and personality. In this paper, we systematically investigate advanced models on the taskoriented dialogue translation task, including sentence-level, document-level and non-autoregressive NMT models. Besides, we explore existing techniques such as data selection, back/forward translation, larger batch learning, finetuning and domain adaptation. To alleviate low-resource problem, we transfer general knowledge from four different pre-training models to the downstream task. Encouragingly, we find that the best model with mBART pre-training pushes the SOTA performance on WMT20 English-German and IWSLT DIA-LOG Chinese-English datasets up to 62.67 and 23.21 BLEU points, respectively. 1
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations鈥揷itations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright 漏 2024 scite LLC. All rights reserved.
Made with 馃挋 for researchers
Part of the Research Solutions Family.