Translation Framework Research Articles

A Study on Automatic Machine Translation Tools: A Comparative Error Analysis Between DeepL and Yandex for Russian-Italian Medical Translation

The present research is aimed at conducting a study with regard to Russian-Italian medical translation on the current development of two Machine Translation tools that feature prominently in today’s Neural Machine Translation framework, namely DeepL and Yandex. For the purpose of our research, we have selected a number of Russian medical articles: three highly specialized and three popular-science articles concerning coronavirus pandemic. Such a choice is justified by the willingness not only to analyse recently published scientific documents but also to investigate the particular linguistic implications of 2020’s coronavirus pandemic outbreak. In fact, during the period of pandemic a set of terms has been introduced and coined in every-day communication and entered the boundaries of scientific terminology. We have considered this existing linguistic phenomenon as a proper condition to test the performances of Machine Translation tools. In particular, we discuss the most relevant features of the comparative error analysis as well as the BLEU metric for both DeepL and Yandex.

Machine Translation System Based on Deep Learning

The translation fusion of multi-machine translation is a strategy that has been proposed very early to improve the quality of machine translation. It has been fully discussed under the framework of statistical machine translation, and further breakthroughs in performance have entered a bottleneck period. Since the fusion methods of existing methods are mostly limited to surface features and lack effective deep fusion strategies, this paper encodes source language sentences and multiple system translations separately to achieve fusion at the coding level. In the specific decoding, we limit the decoding space, so that the translation model can obtain a higher performance improvement on less data. When increasing the values of the two hyper-parameters α and β, the translation effect can be gradually improved. The performance of the model reaches the maximum when a and b are respectively set to 0.4, and the translation performance will decrease if it continues to increase.

Application of the adverse outcome pathway conceptual framework for translation of mechanistic data into regulatory decisions: adverse outcome pathways for kidney injury as case study

Application of the adverse outcome pathway conceptual framework for translation of mechanistic data into regulatory decisions: adverse outcome pathways for kidney injury as case study

Deep Learning Approach for the Morphological Synthesis in Malayalam and Tamil at the Character Level

Morphological synthesis is one of the main components of Machine Translation (MT) frameworks, especially when any one or both of the source and target languages are morphologically rich. Morphological synthesis is the process of combining two words or two morphemes according to the Sandhi rules of the morphologically rich language. Malayalam and Tamil are two languages in India which are morphologically abundant as well as agglutinative. Morphological synthesis of a word in these two languages is challenging basically because of the following reasons: (1) Abundance in morphology; (2) Complex Sandhi rules; (3) The possibilty in Malayalam to form words by combining words that belong to different syntactic categories (for example, noun and verb); and (4) The construction of a sentence by combining multiple words. We formulated the task of the morphological generation of nouns and verbs of Malayalam and Tamil as a character-to-character sequence tagging problem. In this article, we used deep learning architectures like Recurrent Neural Network (RNN) , Long Short-Term Memory Networks (LSTM) , Gated Recurrent Unit (GRU) , and their stacked and bidirectional versions for the implementation of morphological synthesis at the character level. In addition to that, we investigated the performance of the combination of the aforementioned deep learning architectures and the Conditional Random Field (CRF) in the morphological synthesis of nouns and verbs in Malayalam and Tamil. We observed that the addition of CRF to the Bidirectional LSTM/GRU architecture achieved more than 99% accuracy in the morphological synthesis of Malayalam and Tamil nouns and verbs.

A pre-translation framework for public service translation: A sociological approach to enhance translation effectiveness

In this paper a pre-translation framework developed for public service translation is presented. The framework was developed from a sociological standpoint “whereby translation practice can be viewed in relation to people and institutions involved in it or affected by it” (Taibi & Ozolins, 2016, pp. 63-64) with the aim to provide contextualization and understanding that precedes the actual translation stage. After outlining the framework, its usefulness in obtaining key pre-translation information is showcased through an application form in the Spanish immigration context, i.e. an authentic public service text. Lastly, the paper shows how the information gathered based on the framework will assist in proposing an effective English translation of this application form in the future.

The usability and applicability of knowledge translation theories, models, and frameworks for research in the context of a national health service

BackgroundTranslating research findings into service improvements for patients and/or policy changes is a key challenge for health service organizations. The Health Service Executive (HSE) in Ireland launched the Action Plan for Health Research 2019–2029, as reported by Terrés (HSE, Dublin, 2019), one of the goals of which is to maximize the impact of the research that takes place within the service to achieve improvements in patient care, services, or policy change.The purpose of this research is to review the literature on knowledge translation theories, models, and frameworks (TMFs) and to assess the suitability of the TMFs for HSE use, selecting one or more for this purpose. The aim is to produce guidance for HSE researchers and other health services staff, validate the usability of the framework(s) with researchers, and review and implement the guidance. It was hoped that identifying a suitable methodology would provide the means to increase the uptake and application of research findings, and reduce research wastage. This paper reports on the first part of the study: the review, assessment, and selection of knowledge translation TMFs for a national health service.MethodsAn interdisciplinary working group of academic experts in implementation science, research wastage, and knowledge translation, along with key representatives from research funders (Health Research Board) and HSE personnel with expertise in quality improvement and research management, undertook a three-stage review and selection process to identify a knowledge translation TMF that would be suitable and usable for HSE purposes. The process included a literature review, consensus exercise, and a final consensus workshop. The review group adopted the Theory Comparison and Selection Tool (T-CaST) developed by Birken et al. (Implement Sci 13: 143, 2018) to review knowledge translation theories, models, and frameworks.ResultsFrom 247 knowledge translation TMFs initially identified, the first stage of the review identified 18 that met the criteria of validity, applicability, relevance, usability, and ability to be operationalized in the local context. A further review by a subgroup of the working group reduced this number to 11. A whole-group review selected six of these to be reviewed at a facilitated consensus workshop, which identified three that were suitable and applicable for HSE use. These were able to be mapped onto the four components of the HSE knowledge translation process: knowledge creation, knowledge into action, transfer and exchange of knowledge, and implementation and sustainability.ConclusionThe multiplicity of knowledge translation TMFs presents a challenge for health service researchers in making decisions about the appropriate methods for disseminating their research. Building a culture that uses research knowledge and evidence is important for organizations seeking to maximize the benefits from research. Supporting researchers with guidance on how to disseminate and translate their research can increase the uptake and application of research findings.The use of robust selection criteria enabled the HSE to select relevant TMFs and develop a process for increasing the dissemination and translation of research knowledge. The guidance developed to inform and educate researchers and knowledge users is expected to increase organizational capacity to promote a culture of research knowledge and evidence use within the HSE.

Knowledge Translation Framework for Translating the Name of Thai Dishes from Thai into the Chinese Language

Knowledge Translation Framework for Translating the Name of Thai Dishes from Thai into the Chinese Language

DMDIT: Diverse multi-domain image-to-image translation

Cross-domain image translation studies have shown brilliant progress in recent years, which intend to learn the mapping between two different domains. A good cross-domain image translation model should meet the following conditions: (1) do not rely on paired dataset, (2) can deal with multiple domains, (3) obtain diverse outputs with the same source image. Most state-of-art studies are devoted to addressing two of them i.e., either (1) and (2), or (1) and (3). In this paper, we construct a unified diverse multi-domain image to image translation framework (DMDIT) which can satisfy the above three requirements simultaneously. Different from traditional approaches, the proposed generator can achieve diverse and multi-label image-to-image translation while retaining the underlying features of the input image. The diverse outputs are obtained through a latent noise sampled from the normal distribution randomly. To further improve the multiplicity of the outputs, we propose a novel style regularization loss to restrain the latent noise. The mode collapse problem usually occurs due to the lack of constraints on the noise, so we embed a noise separation module in the discriminator to avoid this issue. In addition, we apply an attention mechanism to make the model attentively focus on the most attribute-relevant regions, helping to improve the quality of the generated images. Extensive qualitative and quantitative evaluations clearly demonstrate the effectiveness of our approach.

Systematic review of interventions targeting fundamental care to reduce hospital-associated decline in older patients.

To examine the effectiveness of targeted nursing interventions on mobilization, nutrition and cognitive engagement to reduce functional and hospital-associated decline (HAD) in older patients. Systematic review of experimental studies using randomized and quasi-experimental designs. We searched electronic databases CINAHL, MEDLINE, EMBASE, Cochrane library, google scholar and BMJ quality reports from January 2009 to February 2020. We reviewed intervention studies that targeted ward nursing teams to increase mobilization, nutrition or cognitive engagement of older adults. Inclusion criteria included older patients, acute care (medical, surgical and older adult wards) and reporting patient level outcomes. Quality appraisal included the Joanna Briggs Critical Appraisal Checklist for Quasi-Experimental Studies. From 1729 papers, 18 studies using quasi-experimental and pre-post designs were selected. Study heterogeneity necessitated a narrative synthesis. The quality of evidence was low to moderate. All studies used multicomponent strategies, and 10 studies used evidence translation frameworks to align interventions to local barriers. Overall, 74% (n=14) of studies reported a significant improvement in the stated primary outcome. Eight studies reported a significant increase in mobilization (e.g., sitting in a chair or walking), and four reported improved functional outcomes. Five studies improved nutrition outcomes (e.g., protein or energy intake), and three studies reported a significant reduction in delirium. Acknowledging methodological limitations, the evidence indicates that nursing teams using evidence-translation frameworks can improve mobilization, nutrition and cognitive engagement in acute care settings. Future research requires higher-quality pragmatic trial designs, standardized outcomes, staff co-designed interventions, evidence-translation frameworks and patient engagement to make more confident inference about effectiveness. Nursing teams with the support of hospital management have to address ward and system barriers to prioritize fundamental care to improve patient outcomes. There is sufficient evidence on multicomponent interventions and implementation strategies to inform nurse-led quality improvement.

Associations between Exposure to Organochlorine Chemicals and Endometriosis: A Systematic Review of Experimental Studies and Integration of Epidemiological Evidence.

Background:Growing epidemiological evidence suggests that organochlorine chemicals (OCCs), including 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD), may play a role in the pathogenesis of endometriosis.Objectives:We aimed to systematically review the experimental evidence (in vivo and in vitro) on the associations between exposure to OCCs and endometriosis-related end points.Methods:A systematic review protocol was developed following the National Toxicology Program /Office of Health Assessment and Translation (NTP/OHAT) framework and managed within a web-based interface. In vivo studies designed to evaluate the impact of OCCs on the onset or progression of endometriosis and proliferation of induced endometriotic lesions were eligible. Eligible in vitro studies included single-cell and co-culture models to evaluate the proliferation, migration, and/or invasion of endometrial cells. We applied the search strings to PubMed, Web of Science, and Scopus®. A final search was performed on 24 June 2020. Assessment of risk of bias and the level of evidence and integration of preevaluated epidemiological evidence was conducted using NTP/OHAT frameworkResults:Out of 812 total studies, 39 met the predetermined eligibility criteria (15 in vivo, 23 in vitro, and 1 both). Most studies () tested TCDD and other dioxin-like chemicals. In vivo evidence supported TCDD’s promotion of endometriosis onset and lesion growth. In vitro evidence supported TCDD’s promotion of cell migration and invasion, but there was insufficient evidence for cell proliferation. In vitro evidence further supported the roles of the aryl hydrocarbon receptor and matrix metalloproteinases in mediating steroidogenic disruption and inflammatory responses. Estrogen interactions were found across studies and end points.Conclusion:Based on the integration of a high level of animal evidence with a moderate level of epidemiological evidence, we concluded that TCDD was a known hazard for endometriosis in humans and the conclusion is supported by mechanistic in vitro evidence. Nonetheless, there is need for further research to fill in our gaps in understanding of the relationship between OCCs and their mixtures and endometriosis, beyond the prototypical TCDD. https://doi.org/10.1289/EHP8421

A Binary Translation Framework for Automated Hardware Generation

As applications move to the edge, efficiency in computing power and power/energy consumption is required. Heterogeneous computing promises to meet these requirements through application-specific hardware accelerators. Runtime adaptivity might be of paramount importance to realize the potential of hardware specialization, but further study is required on workload retargeting and offloading to reconfigurable hardware. This article presents our framework for the exploration of both offloading and hardware generation techniques. The framework is currently able to process instruction sequences from MicroBlaze, ARMv8, and riscv32imaf binaries, and to represent them as Control and Dataflow Graphs for transformation to implementations of hardware modules. We illustrate the framework's capabilities for identifying binary sequences for hardware translation with a set of 13 benchmarks.

TERMS OF THE MENTAL LEXICON IN THE PROFESSIONAL AVIATION LANGUAGE TRANSLATORS

The present paper establishes the possibility of presenting the mental lexicon of a technical translator in the form of a classification frame that structures the terminological composition of scientific and technical language in the translation consciousness during the cognition of aviation reality. It is stated that the terms, which are the main means of fixation, accumulation and transfer of scientific knowledge, become the object of cognitive activity of the translator, who performs cognitive work on their accumulation and structuring in their own mental space to operate them during translation, which emphasizes the heuristic nature translation activities. At the same time, the linguistic component of the mental lexicon allows us to consider its structure, starting from the language itself. The paper attempts to present a mental lexicon within the framework of scientific and technical translation in the form of a set of frames that arrange the terminological structure of language in the mind of the translator as they learn the aviation reality. It is noted that in this case the main task is to show the possibilities of the frame in terms of accumulating and structuring the verbal knowledge about a particular aviation phenomenon or process, in English and Ukrainian with simultaneous identification of interlanguage equivalents.

DDAT: Dual domain adaptive translation for low-resolution face verification in the wild

Low-resolution (LR) face verification has received much attention because of its wide applicability in real scenarios, especially in long-distance surveillance. However, the poor quality and scarcity of training data make the accuracy far from satisfactory. In this paper, we propose an end-to-end LR face translation and verification framework to improve the generation quality of face images and face verification accuracy simultaneously. We design a dual domain adaptive structure to generate high-quality images. On one hand, the structure can reduce the domain gap between training data and test data. On the other hand, the structure preserves identity consistency and low-level attributes. Meanwhile, in order to make the whole model more robust, we treat the generated images of the target domain as an extension of the training data. We conduct extensive comparative experiments on multiple benchmark data sets. Experimental results verify that our method achieves improved results in high-quality face generation and LR face verification. In particular, our model DDAT reduces FID to 18.63 and 39.55 on the source and the target domain from 254.7 and 206.19 of the up-sampling results, respectively. Our method outperforms competing approaches by more than 10 percentage points in terms of face verification accuracy on multiple surveillance benchmarks.

Identification of knowledge translation theories, models or frameworks suitable for health technology reassessment: a survey of international experts

ObjectiveHealth technology reassessment (HTR) is a field focused on managing a technology throughout its life cycle for optimal use. The process results in one of four possible recommendations: increase use,...

General Translation and Covid-19 Terminologies

This article titled ‘General Translation and its texts', is based on the notion of general translation, the composition of its texts and some COVID19 terminologies as a practical example of General Translation especially in the traditional African language, example of Khana. This communication wants to examine general translation and its texts in the pragmatic framework of translation. Study deals with reevaluating the different information misinterpreted in the field of General Translation and its Texts and thus the difference between the concept of general translation and specialized translation within the framework of pragmatic translation. The article then wants to re-establish critical difference in their specificities appropriate to this class of translation with some illustrative documents that we have chosen. The methodology of this study will be documentary hence involves published journals stored in the libraries and on the internet, carrying relevant information to our work. This article offers definitions and proper classification to general translation and its texts in the context of pragmatic translation, because in the study of translation, the notion of general translation and its text is not well structured as a category of translation among the genres and genres of texts in translation and especially in the treatment of translation typologies and text typologies.

Eco-translatology-based Analysis of Children’s Literature Translation—A Case Study: Peter Pan

Children’s literature occupies a peripheral position in literature system according to the polysystem theory so that the translators of children’s literature can manipulate the texts with great liberty. The translator of children’s literature in the ternary relation of translation, namely the source texts, the translator and the target text, is in a relatively important position. Thus, it is a feasible way to analyze the translation of children’s literature from the translator-centered perspective. Eco-translatology is a translator-centered translation theory, aiming to analyze how the translator selects and adapts during the translation process in the translational eco-environment. In this paper, the author will adopt Eco-translatology as the translation framework to analyze the translation of children’s literature, and try to explore how ‘children’, an important factor in the translational eco-environment, influences the translator’s selection and adaptation in the process of translating children’s literature. Furthermore, the author will take Peter Pan as a case study, comparing two Chinese versions of this book to analyze how the two translators adapt and select differently from those three dimensions during the translation process, as one follows the target-reader-oriented strategy and the other one follows the source-text-oriented strategy.

Modality Translation in Remote Sensing Time Series

Modality translation, which aims to translate images from a source modality to a target one, has attracted a growing interest in the field of remote sensing recently. Compared to translation problems in multimedia applications, modality translation in remote sensing often suffers from inherent ambiguities, i.e., a single input image could correspond to multiple possible outputs, and the results may not be valid in the following image interpretation tasks, such as classification and change detection. To address these issues, we make the attempt to utilizing time-series data to resolve the ambiguities. We propose a novel multimodality image translation framework, which exploits temporal information from two aspects: 1) by introducing a guidance image from given temporally neighboring images in the target modality, we employ a feature mask module and transfer semantic information from temporal images to the output without requiring the use of any semantic labels and 2) while incorporating multiple pairs of images in time series, a temporal constraint is formulated during the learning process in order to guarantee the uniqueness of the prediction result. We also build a multimodal and multitemporal dataset that contains synthetic aperture radar (SAR), visible, and short-wave length infrared band (SWIR) image time series of the same scene to encourage and promote research on modality translation in remote sensing. Experiments are conducted on the dataset for two cross-modality translation tasks (SAR to visible and visible to SWIR). Both qualitative and quantitative results demonstrate the effectiveness and superiority of the proposed model.

Towards Fully Automated Manga Translation

We tackle the problem of machine translation of manga, Japanese comics. Manga translation involves two important problems in machine translation: context-aware and multimodal translation. Since text and images are mixed up in an unstructured fashion in Manga, obtaining context from the image is essential for manga translation. However, it is still an open problem how to extract context from image and integrate into MT models. In addition, corpus and benchmarks to train and evaluate such model is currently unavailable. In this paper, we make the following four contributions that establishes the foundation of manga translation research. First, we propose multimodal context-aware translation framework. We are the first to incorporate context information obtained from manga image. It enables us to translate texts in speech bubbles that cannot be translated without using context information (e.g., texts in other speech bubbles, gender of speakers, etc.). Second, for training the model, we propose the approach to automatic corpus construction from pairs of original manga and their translations, by which large parallel corpus can be constructed without any manual labeling. Third, we created a new benchmark to evaluate manga translation. Finally, on top of our proposed methods, we devised a first compleheisive system for fully automated manga translation.

Efficient English Translation Method and Analysis Based on the Hybrid Neural Network

Neural machine translation has been widely concerned in recent years. The traditional sequential neural network framework of English translation has obvious disadvantages because of its poor ability to capture long-distance information, and the current improved framework, such as the recurrent neural network, still cannot solve this problem very well. In this paper, we propose a hybrid neural network that combines the convolutional neural network (CNN) and long short-term memory (LSTM) and introduce the attention mechanism based on the encoder-decoder structure to improve the translation accuracy, especially for long sentences. In the experiment, this model is implemented based on TensorFlow, and the results show that the BLEU value of the proposed method is obviously improved compared with the traditional machine learning model, which proves the effectiveness of our method in English-Chinese translation.

Program evaluation of GLA:D® Australia: Physiotherapist training outcomes and effectiveness of implementation for people with knee osteoarthritis

ObjectiveEvaluate the implementation of Good Life with osteoArthritis from Denmark (GLA:D®) for knee osteoarthritis in Australia using the RE-AIM QuEST (Reach, Effectiveness, Adoption, Implementation, Maintenance Qualitative Evaluation for Systematic Translation) framework. DesignPhysiotherapists completed surveys before and after GLA:D® training (2017–2020) to assess practices, and barriers and enablers to implementation. Patients completed online baseline, 3-month (post-treatment) and 12-month patient reported outcomes. Effective implementation was defined as within-participant moderate effect size (ES, ≥0.50) for average pain (100 ​mm VAS) and Knee Injury and Osteoarthritis Outcome Score quality of life scores (KOOS-QoL), and small effect size (≥0.20) for health-related quality of life (EQ-5D-5L). ResultsReach: 1064 physiotherapists and 1945 patients from all states and territories participated. Key barriers included out-of-pocket cost to patients, and program suitability for culturally and linguistically diverse communities. Effectiveness: Following training, more physiotherapists reported discussing treatment goals and weight management, and prescribing supervised, neuromuscular exercise. Patient outcomes at 3- and 12-months (n ​= ​1044 [54%] and 927 [48%]) reflected effective implementation, including reduced pain (ES, 95%CI ​= ​0.72, 0.62–0.84; and 0.65, 0.54–0.77) and improved KOOS-QoL (0.79, 0.69–0.90; and 0.93, 0.81–1.04) and EQ-5D-5L (0.43, 0.31–0.54; and 0.46, 0.35–0.58) scores. Adoption: 297 sites (264 private, 33 public) implemented GLA:D®. Implementation: Most patients completed at least one education (90%), and 10 exercise-therapy (78%) sessions. Adequate staffing to support program delivery was a key enabler. Maintenance: 99% of sites (293/297) continued offering the program in July 2020. ConclusionTraining changed practice and was associated with effective widespread implementation of GLA:D® in Australia.