SemMT: A Semantic-based Testing Approach for Machine Translation Systems
The Hong Kong University of Science and Technology（香港科技大学）；Institute of Software, Chinese Academy of Sciences（中国科学院软件研究所）；University of Chinese Academy of Sciences（中国科学院大学）；Huazhong University of Science and Technology（华中科技大学）
Jialun Cao, Meiziniu Li, Yeting Li, Ming Wen, Shing-Chi Cheung
Machine translation has wide applications in daily life. In mission-critical applications such as translating official documents, incorrect translation can have unpleasant or sometimes catastrophic consequences. This motivates recent research on testing methodologies for machine translation systems. Existing methodologies mostly rely on metamorphic relations designed at the textual level (e.g., Levenshtein distance) or syntactic level (e.g., the distance between grammar structures) to determine the correctness of translation results. However, these metamorphic relations do not consider whether the original and translated sentences have the same meaning (i.e., Semantic similarity). Therefore, in this paper, we propose SemMT, an automatic testing approach for machine translation systems based on semantic similarity checking. SemMT applies round-trip translation and measures the semantic similarity between the original and translated sentences. Our insight is that the semantics expressed by the logic and numeric constraint in sentences can be captured using regular expressions (or deterministic finite automata) where efficient equivalence/similarity checking algorithms are available. Leveraging the insight, we propose three semantic similarity metrics and implement them in SemMT. The experiment result reveals SemMT can achieve higher effectiveness compared with state-of-the-art works, achieving an increase of 21% and 23% on accuracy and F-Score, respectively. We also explore potential improvements that can be achieved when proper combinations of metrics are adopted. Finally, we discuss a solution to locate the suspicious trip in round-trip translation, which may shed lights on further exploration.
In recent years, machine translation software has increasingly been integrated into our daily lives. People routinely use machine translation for various applications, such as describing symptoms to a foreign doctor and reading political news in a foreign language. However, the complexity and intractability of neural machine translation (NMT) models that power modern machine translation make the robustness of these systems difficult to even assess, much less guarantee. Machine translation systems can return inferior results that lead to misunderstanding, medical misdiagnoses, threats to person…
Machine translation software has seen rapid progress in recent years due to the advancement of deep neural networks. People routinely use machine translation software in their daily lives, such as ordering food in a foreign restaurant, receiving medical diagnosis and treatment from foreign doctors, and reading international political news online. However, due to the complexity and intractability of the underlying neural networks, modern machine translation software is still far from robust and can produce poor or incorrect translations; this can lead to misunderstanding, financial loss, threat…
Tencent Inc.（腾讯公司）；University of Illinois at Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校）；University of Texas at Dallas（德克萨斯大学达拉斯分校）；ETH Zurich（苏黎世联邦理工学院）
Neural Machine Translation (NMT) has been widely adopted recently due to its advantages compared with the traditional Statistical Machine Translation (SMT). However, an NMT system still often produces translation failures due to the complexity of natural language and sophistication in designing neural networks. While in-house black-box system testing based on reference translations (i.e., examples of valid translations) has been a common practice for NMT quality assurance, an increasingly critical industrial practice, named in-vivo testing, exposes unseen types or instances of translation fail…
In the current work, we explore the enrichment in the machine translation output when the training parallel corpus is augmented with the introduction of sentiment analysis. The paper discusses the preparation of the same sentiment tagged English-Bengali parallel corpus. The preparation of raw parallel corpus, sentiment analysis of the sentences and the training of a Character Based Neural Machine Translation model using the same has been discussed extensively in this paper. The output of the translation model has been compared with a base-line translation model using automated metrics such as …
Machine Translation Testing via Syntactic Tree Pruning
State Key Laboratory for Novel Software Technology, Nanjing University（南京大学计算机软件新技术国家重点实验室）；Manning College of Information & Computer Sciences, University of Massachusetts Amherst（马萨诸塞大学阿默斯特分校曼宁信息与计算机科学学院）
本文提出基于语法树剪枝（STP）的蜕变测试方法，通过保留核心语义的剪枝策略生成源句子对，有效发现Google Translate和Bing Microsoft Translator中的大量独特翻译错误，精度和召回率显著优于现有技术。
University of Melbourne（墨尔本大学）；Facebook AI（Facebook人工智能研究院）；Twitter Cortex
Mistranslated numbers have the potential to cause serious effects, such as financial loss or medical misinformation. In this work we develop comprehensive assessments of the robustness of neural machine translation systems to numerical text via behavioural testing. We explore a variety of numerical translation capabilities a system is expected to exhibit and design effective test examples to expose system underperformance. We find that numerical mistranslation is a general issue: major commercial systems and state-of-the-art research models fail on many of our test examples, for high- and low-…
The training data used in NMT is rarely controlled with respect to specific attributes, such as word casing or gender, which can cause errors in translations. We argue that predicting the target word and attributes simultaneously is an effective way to ensure that translations are more faithful to the training data distribution with respect to these attributes. Experimental results on two tasks, uppercased input translation and gender prediction, show that this strategy helps mirror the training data distribution in testing. It also facilitates data augmentation on the task of uppercased input…
Traditional machine translation (MT) metrics provide an average measure of translation quality that is insensitive to the long tail of behavioral problems in MT. Examples include translation of numbers, physical units, dropped content and hallucinations. These errors, which occur rarely and unpredictably in Neural Machine Translation (NMT), greatly undermine the reliability of state-of-the-art MT systems. Consequently, it is important to have visibility into these problems during model development. Towards this direction, we introduce SALTED, a specifications-based framework for behavioral tes…
Technical University of Munich（慕尼黑工业大学）；University of Zurich（苏黎世大学）；ETH Zurich（苏黎世联邦理工学院）
We explore the suitability of self-attention models for character-level neural machine translation. We test the standard transformer model, as well as a novel variant in which the encoder block combines information from nearby characters using convolutions. We perform extensive experiments on WMT and UN datasets, testing both bilingual and multilingual translation to English using up to three input languages (French, Spanish, and Chinese). Our transformer variant consistently outperforms the standard transformer at the character-level and converges faster while learning more robust character-l…
In order to control computational complexity, neural machine translation (NMT) systems convert all rare words outside the vocabulary into a single unk symbol. Previous solution (Luong et al., 2015) resorts to use multiple numbered unks to learn the correspondence between source and target rare words. However, testing words unseen in the training corpus cannot be handled by this method. And it also suffers from the noisy word alignment. In this paper, we focus on a major type of rare words -- named entity (NE), and propose to translate them with character level sequence to sequence model. The N…
Neural Machine Translation (NMT) systems are known to degrade when confronted with noisy data, especially when the system is trained only on clean data. In this paper, we show that augmenting training data with sentences containing artificially-introduced grammatical errors can make the system more robust to such errors. In combination with an automatic grammar error correction system, we can recover 1.5 BLEU out of 2.4 BLEU lost due to grammatical errors. We also present a set of Spanish translations of the JFLEG grammar error correction corpus, which allows for testing NMT robustness to real…
National Institute of Information and Communications Technology（日本国立信息通信技术研究所）
This paper presents the first large-scale meta-evaluation of machine translation (MT). We annotated MT evaluations conducted in 769 research papers published from 2010 to 2020. Our study shows that practices for automatic MT evaluation have dramatically changed during the past decade and follow concerning trends. An increasing number of MT evaluations exclusively rely on differences between BLEU scores to draw conclusions, without performing any kind of statistical significance testing nor human evaluation, while at least 108 metrics claiming to be better than BLEU have been proposed. MT evalu…
Polish Japanese Institute of Technology（波兰日本信息技术学院）
This research explores the effects of various training settings on a Polish to English Statistical Machine Translation system for spoken language. Various elements of the TED, Europarl, and OPUS parallel text corpora were used as the basis for training of language models, for development, tuning and testing of the translation system. The BLEU, NIST, METEOR and TER metrics were used to evaluate the effects of the data preparations on the translation results.
In this research, we have established, through empirical testing, a law that relates the number of translating hops to translation accuracy in sequential machine translation in Google Translate. Both accuracy and size decrease with the number of hops; the former displays a decrease closely following a power law. Such a law allows one to predict the behavior of translation chains that may be built as society increasingly depends on automated devices.
A confidence measure is able to estimate the reliability of an hypothesis provided by a machine translation system. The problem of confidence measure can be seen as a process of testing : we want to decide whether the most probable sequence of words provided by the machine translation system is correct or not. In the following we describe several original word-level confidence measures for machine translation, based on mutual information, n-gram language model and lexical features language model. We evaluate how well they perform individually or together, and show that using a combination of c…
Facebook AI Research（脸书人工智能研究院）；Paris-Dauphine University（巴黎第九大学）；University College London（伦敦大学学院）；Facebook（脸书）
With little to no parallel data available for programming languages, unsupervised methods are well-suited to source code translation. However, the majority of unsupervised machine translation approaches rely on back-translation, a method developed in the context of natural language translation and one that inherently involves training on noisy inputs. Unfortunately, source code is highly sensitive to small changes; a single token can result in compilation failures or erroneous programs, unlike natural languages where small inaccuracies may not change the meaning of a sentence. To address this …
Oregon State University（俄勒冈州立大学）；Baidu Research（百度研究院）
Beam search optimization resolves many issues in neural machine translation. However, this method lacks principled stopping criteria and does not learn how to stop during training, and the model naturally prefers the longer hypotheses during the testing time in practice since they use the raw score instead of the probability-based score. We propose a novel ranking method which enables an optimal beam search stopping criteria. We further introduce a structured prediction loss function which penalizes suboptimal finished candidates produced by beam search during training. Experiments of neural m…
Machine translation systems are very sensitive to the domains they were trained on. Several domain adaptation techniques have been deeply studied. We propose a new technique for neural machine translation (NMT) that we call domain control which is performed at runtime using a unique neural network covering multiple domains. The presented approach shows quality improvements when compared to dedicated domains translating on any of the covered domains and even on out-of-domain data. In addition, model parameters do not need to be re-estimated for each domain, making this effective to real use cas…
This paper presents Centre for Development of Advanced Computing Mumbai's (CDACM) submission to NLP Tools Contest on Statistical Machine Translation in Indian Languages (ILSMT) 2015 (collocated with ICON 2015). The aim of the contest was to collectively explore the effectiveness of Statistical Machine Translation (SMT) while translating within Indian languages and between English and Indian languages. In this paper, we report our work on all five language pairs, namely Bengali-Hindi (\bnhi), Marathi-Hindi (\mrhi), Tamil-Hindi (\tahi), Telugu-Hindi (\tehi), and English-Hindi (\enhi) for Health,…
Polish Japanese Academy of Information Technology（波兰-日本信息技术学院）
This research explores effects of various training settings between Polish and English Statistical Machine Translation systems for spoken language. Various elements of the TED parallel text corpora for the IWSLT 2014 evaluation campaign were used as the basis for training of language models, and for development, tuning and testing of the translation system as well as Wikipedia based comparable corpora prepared by us. The BLEU, NIST, METEOR and TER metrics were used to evaluate the effects of data preparations on translation results. Our experiments included systems, which use lemma and morphol…
We explore zero-shot adaptation, where a general-domain model has access to customer or domain specific parallel data at inference time, but not during training. We build on the idea of Retrieval Augmented Translation (RAT) where top-k in-domain fuzzy matches are found for the source sentence, and target-language translations of those fuzzy-matched sentences are provided to the translation model at inference time. We propose a novel architecture to control interactions between a source sentence and the top-k fuzzy target-language matches, and compare it to architectures from prior work. We con…
We present a challenge set for French --> English machine translation based on the approach introduced in Isabelle, Cherry and Foster (EMNLP 2017). Such challenge sets are made up of sentences that are expected to be relatively difficult for machines to translate correctly because their most straightforward translations tend to be linguistically divergent. We present here a set of 506 manually constructed French sentences, 307 of which are targeted to the same kinds of structural divergences as in the paper mentioned above. The remaining 199 sentences are designed to test the ability of the sy…
Primer Technologies Inc.（Primer Technologies公司）；Technische Universität Berlin（柏林工业大学）
The creation of a quality summarization dataset is an expensive, time-consuming effort, requiring the production and evaluation of summaries by both trained humans and machines. If such effort is made in one language, it would be beneficial to be able to use it in other languages without repeating human annotations. To investigate how much we can trust machine translation of such a dataset, we translate the English dataset SummEval to seven languages and compare performance across automatic evaluation measures. We explore equivalence testing as the appropriate statistical paradigm for evaluati…
Word and phrase tables are key inputs to machine translations, but costly to produce. New unsupervised learning methods represent words and phrases in a high-dimensional vector space, and these monolingual embeddings have been shown to encode syntactic and semantic relationships between language elements. The information captured by these embeddings can be exploited for bilingual translation by learning a transformation matrix that allows matching relative positions across two monolingual vector spaces. This method aims to identify high-quality candidates for word and phrase translation more c…
We introduce the cross-match test - an exact, distribution free, high-dimensional hypothesis test as an intrinsic evaluation metric for word embeddings. We show that cross-match is an effective means of measuring distributional similarity between different vector representations and of evaluating the statistical significance of different vector embedding models. Additionally, we find that cross-match can be used to provide a quantitative measure of linguistic similarity for selecting bridge languages for machine translation. We demonstrate that the results of the hypothesis test align with our…
Non-autoregressive approaches aim to improve the inference speed of translation models by only requiring a single forward pass to generate the output sequence instead of iteratively producing each predicted token. Consequently, their translation quality still tends to be inferior to their autoregressive counterparts due to several issues involving output token interdependence. In this work, we take a step back and revisit several techniques that have been proposed for improving non-autoregressive translation models and compare their combined translation quality and speed implications under thi…
Polish Japanese Institute of Information Technology（波兰日本信息技术研究所）
This research explores the effects of various training settings from Polish to English Statistical Machine Translation system for spoken language. Various elements of the TED parallel text corpora for the IWSLT 2013 evaluation campaign were used as the basis for training of language models, and for development, tuning and testing of the translation system. The BLEU, NIST, METEOR and TER metrics were used to evaluate the effects of data preparations on translation results. Our experiments included systems, which use stems and morphological information on Polish words. We also conducted a deep a…
Institute of Computing Technology, Chinese Academy of Sciences（中国科学院计算技术研究所）；University of Chinese Academy of Sciences（中国科学院大学）；Sogou Inc.（搜狗公司）；Tsinghua University（清华大学）；Huawei Noah’s Ark Lab, Huawei Technologies（华为技术有限公司华为诺亚方舟实验室）
Although neural machine translation (NMT) has achieved impressive progress recently, it is usually trained on the clean parallel data set and hence cannot work well when the input sentence is the production of the automatic speech recognition (ASR) system due to the enormous errors in the source. To solve this problem, we propose a simple but effective method to improve the robustness of NMT in the case of speech translation. We simulate the noise existing in the realistic output of the ASR system and inject them into the clean parallel data so that NMT can work under similar word distribution…
In this paper, we propose Neural Phrase-to-Phrase Machine Translation (NP$^2$MT). Our model uses a phrase attention mechanism to discover relevant input (source) segments that are used by a decoder to generate output (target) phrases. We also design an efficient dynamic programming algorithm to decode segments that allows the model to be trained faster than the existing neural phrase-based machine translation method by Huang et al. (2018). Furthermore, our method can naturally integrate with external phrase dictionaries during decoding. Empirical experiments show that our method achieves compa…
Polish Japanese Institute of Information Technology（波兰日本信息技术学院）
This new research explores the effects of various training methods on a Polish to English Statistical Machine Translation system for medical texts. Various elements of the EMEA parallel text corpora from the OPUS project were used as the basis for training of phrase tables and language models and for development, tuning and testing of the translation system. The BLEU, NIST, METEOR, RIBES and TER metrics have been used to evaluate the effects of various system and data preparations on translation results. Our experiments included systems that used POS tagging, factored phrase models, hierarchic…
Singapore University of Technology and Design（新加坡科技设计大学）
In neural machine translation, cross entropy (CE) is the standard loss function in two training methods of auto-regressive models, i.e., teacher forcing and scheduled sampling. In this paper, we propose mixed cross entropy loss (mixed CE) as a substitute for CE in both training approaches. In teacher forcing, the model trained with CE regards the translation problem as a one-to-one mapping process, while in mixed CE this process can be relaxed to one-to-many. In scheduled sampling, we show that mixed CE has the potential to encourage the training and testing behaviours to be similar to each ot…
Polish Japanese Institute of Information Technology（波兰日本信息技术学院）
This research investigates the Statistical Machine Translation approaches to translate speech in real time automatically. Such systems can be used in a pipeline with speech recognition and synthesis software in order to produce a real-time voice communication system between foreigners. We obtained three main data sets from spoken proceedings that represent three different types of human speech. TED, Europarl, and OPUS parallel text corpora were used as the basis for training of language models, for developmental tuning and testing of the translation system. We also conducted experiments involv…
Unsupervised machine translation (MT) has recently achieved impressive results with monolingual corpora only. However, it is still challenging to associate source-target sentences in the latent space. As people speak different languages biologically share similar visual systems, the potential of achieving better alignment through visual content is promising yet under-explored in unsupervised multimodal MT (MMT). In this paper, we investigate how to utilize visual content for disambiguation and promoting latent space alignment in unsupervised MMT. Our model employs multimodal back-translation a…
Medical University of Warsaw（华沙医科大学）；Polish Telemedicine Society（波兰远程医疗协会）
Machine translation is evolving quite rapidly in terms of quality. Nowadays, we have several machine translation systems available in the web, which provide reasonable translations. However, these systems are not perfect, and their quality may decrease in some specific domains. This paper examines the effects of different training methods when it comes to Polish - English Statistical Machine Translation system used for the medical data. Numerous elements of the EMEA parallel text corpora and not related OPUS Open Subtitles project were used as the ground for creation of phrase tables and diffe…
Distinguishing Translations by Human, NMT, and ChatGPT: A Linguistic and Statistical Approach
School of Foreign Languages, Shanghai Jiao Tong University（上海交通大学外国语学院）；Department of Computer Science and Engineering, Shanghai Jiao Tong University（上海交通大学计算机科学与工程系）；Institute of Corpus Studies and Applications, Shanghai International Studies University（上海外国语大学语料库研究与应用研究院）
The goal of this work is to design a machine translation (MT) system for a low-resource family of dialects, collectively known as Swiss German, which are widely spoken in Switzerland but seldom written. We collected a significant number of parallel written resources to start with, up to a total of about 60k words. Moreover, we identified several other promising data sources for Swiss German. Then, we designed and compared three strategies for normalizing Swiss German input in order to address the regional diversity. We found that character-based neural MT was the best solution for text normali…
Chhotubhai Gopalbhai Patel Institute of Technology, Uka Tarsadia University（乌卡·塔萨迪亚大学乔图巴伊·戈帕尔巴伊·帕特尔理工学院）；Babumadhav Institute of Information Technology, Uka Tarsadia University（乌卡·塔萨迪亚大学巴布马德哈夫信息技术学院）
Deep Learning techniques are powerful in mimicking humans in a particular set of problems. They have achieved a remarkable performance in complex learning tasks. Deep learning inspired Neural Machine Translation (NMT) is a proficient technique that outperforms traditional machine translation. Performing machine-aided translation on Indic languages has always been a challenging task considering their rich and diverse grammar. The neural machine translation has shown quality results compared to the traditional machine translation approaches. The fully automatic machine translation becomes proble…
We present the task of PreQuEL, Pre-(Quality-Estimation) Learning. A PreQuEL system predicts how well a given sentence will be translated, without recourse to the actual translation, thus eschewing unnecessary resource allocation when translation quality is bound to be low. PreQuEL can be defined relative to a given MT system (e.g., some industry service) or generally relative to the state-of-the-art. From a theoretical perspective, PreQuEL places the focus on the source text, tracing properties, possibly linguistic features, that make a sentence harder to machine translate. We develop a basel…
This paper presents TransRepair, a fully automatic approach for testing and repairing the consistency of machine translation systems. TransRepair combines mutation with metamorphic testing to detect inconsistency bugs (without access to human oracles). It then adopts probability-reference or cross-reference to post-process the translations, in a grey-box or black-box manner, to repair the inconsistencies. Our evaluation on two state-of-the-art translators, Google Translate and Transformer, indicates that TransRepair has a high precision (99%) on generating input pairs with consistent translati…
Universitat d’Alacant（阿利坎特大学）；Univ. of Sheffield（谢菲尔德大学）；Univ. Edinburgh（爱丁堡大学）
A popular application of machine translation (MT) is gisting: MT is consumed as is to make sense of text in a foreign language. Evaluation of the usefulness of MT for gisting is surprisingly uncommon. The classical method uses reading comprehension questionnaires (RCQ), in which informants are asked to answer professionally-written questions in their language about a foreign text that has been machine-translated into their language. Recently, gap-filling (GF), a form of cloze testing, has been proposed as a cheaper alternative to RCQ. In GF, certain words are removed from reference translation…
We propose an approach to build a neural machine translation system with no supervised resources (i.e., no parallel corpora) using multimodal embedded representation over texts and images. Based on the assumption that text documents are often likely to be described with other multimedia information (e.g., images) somewhat related to the content, we try to indirectly estimate the relevance between two languages. Using multimedia as the "pivot", we project all modalities into one common hidden space where samples belonging to similar semantic concepts should come close to each other, whatever th…
NLP Ghana（NLP加纳）；Algorine；Leuphana University Luneburg（吕讷堡洛伊法纳大学）；University of Cape Coast（开普海岸大学）；Edinburgh Napier University（爱丁堡龙比亚大学）；Accra Institute of Technology（阿克拉理工学院）；Tulane University（杜兰大学）；University of Tromso（特罗姆瑟大学）；AiMlCamp；University of Strathclyde（斯特拉斯克莱德大学）；Azubi Africa；Ulster University（阿尔斯特大学）；Centre for Research, Data Science and IT Solutions（研究、数据科学与IT解决方案中心）；University of Energy and Natural Resources（能源与自然资源大学）；Integrated Geospatial Intelligence Application Centre（综合地理空间情报应用中心）；SRH Berlin University of Applied Science（SRH柏林应用科学大学）
We present a parallel machine translation training corpus for English and Akuapem Twi of 25,421 sentence pairs. We used a transformer-based translator to generate initial translations in Akuapem Twi, which were later verified and corrected where necessary by native speakers to eliminate any occurrence of translationese. In addition, 697 higher quality crowd-sourced sentences are provided for use as an evaluation set for downstream Natural Language Processing (NLP) tasks. The typical use case for the larger human-verified dataset is for further training of machine translation models in Akuapem …
