Search

Training supervised video captioning model requires coupled video-caption pairs. However, for many targeted languages, sufficient paired data are not available. To this end, we introduce the unpaired video captioning task aiming to train models without coupled video-caption pairs in target language. To solve the task, a natural choice is to employ a two-step pipeline system: first utilizing video-to-pivot captioning model to generate captions in pivot language and then utilizing pivot-to-target translation model to translate the pivot captions to the target language. However, in such a pipeline system, 1) visual information cannot reach the translation model, generating visual irrelevant target captions; 2) the errors in the generated pivot captions will be propagated to the translation model, resulting in disfluent target captions. To address these problems, we propose the Unpaired Video Captioning with Visual Injection system (UVC-VI). UVC-VI first introduces the Visual Injection Module (VIM), which aligns source visual and target language domains to inject the source visual information into the target language domain. Meanwhile, VIM directly connects the encoder of the video-to-pivot model and the decoder of the pivot-to-target model, allowing end-to-end inference by completely skipping the generation of pivot captions. To enhance the cross-modality injection of the VIM, UVC-VI further introduces a pluggable video encoder, i.e., Multimodal Collaborative Encoder (MCE). The experiments show that UVC-VI outperforms pipeline systems and exceeds several supervised systems. Furthermore, equipping existing supervised systems with our MCE can achieve 4% and 7% relative margins on the CIDEr scores to current state-of-the-art models on the benchmark MSVD and MSR-VTT datasets, respectively., Published at IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI)

Recently, more data-driven approaches are demanding multilingual parallel resources primarily in the cross-language studies. To meet these demands, building multilingual parallel corpora are becoming the focus of many Natural Language Processing (NLP) scientific groups. Unlike monolingual corpora, the number of available multilingual parallel corpora is limited. In this paper, the MulTed, a corpus of subtitles extracted from TEDx talks is introduced. It is multilingual, Part of Speech (PoS) tagged, and bilingually sentence-aligned with English as a pivot language. This corpus is designed for many NLP applications, where the sentence-alignment, the PoS tagging, and the size of corpora are influential such as statistical machine translation, language recognition, and bilingual dictionary generation. Currently, the corpus has subtitles that cover 1100 talks available in over 100 languages. The subtitles are classified based on a variety of topics such as Business, Education, and Sport. Regarding the PoS tagging, the Treetagger, a language-independent PoS tagger, is used; then, to make the PoS tagging maximally useful, a mapping process to a universal common tagset is performed. Finally, we believe that making the MulTed corpus available for a public use can be a significant contribution to the literature of NLP and corpus linguistics, especially for under-resourced languages.

Multi-lingual question generation (QG) is the task of generating natural language questions for single answer passage in any given language. In this paper, we design a system for supporting multi-lingual QG in the "People Also Ask" (PAA) module for Bing. For zero shot setting, the primary challenge is to transfer the knowledge from trained QG model in the pivot language to other languages without further addition of training data in these languages. Compared to other zero-shot tasks, the differentiating and challenging aspect in QG is to preserve the question structure so that the resulting output is interrogative. Existing models for similar tasks tend to generate natural language queries or copy sub-span of the passage, failing to preserve the question structure. In our work, we demonstrate how knowledge transfer in multi-lingual IQG (Interrogative QG) can be significantly improved using auxiliary tasks either in multi-task or pre-training task setting. We explore two kinds of tasks - cross-lingual translation and multi-lingual denoising auto-encoding of questions, especially when using translate-train. Using data for 13 languages from Bing PAA as well as online A/B tests, we show that both of these tasks significantly improve the quality of zero-shot IQG on non-trained languages.

This paper is an attempt to focus on investigating the pivot (bridge) language technique, where the pivot language used to improve Statistical Machine Translation (SMT) quality. In this case, Indonesian is used as a pivot language, where each available corpus can be used to support the Madurese-Sundanese language pair. Experiments that have been carried out using the parallel corpus of the Indonesian-Madurese and Indonesian-Sundanese languages ​​are equal to 5K and 6K sentences respectively, while the monolingual corpus used Malay, Sundanese and Indonesian each at 10K, 10K and 100K sentences. This study compares the results of applying the Triangulation and Transfer methods using Indonesian as a pivot language. The results of the research proved that the Triangulation method has better acceleration when compared to the Transfer method. From the results of the experiments conducted, the Triangulation method increased the average Indonesian pivot-based SMT testing by 6.18% for Madura-Sundanese SMT and 7.27% for Madurese-Sundanese SMT., Makalah ini merupakan upaya untuk fokus pada investigasi teknik bahasa pivot (penghubung) , di mana bahasa penghubung digunakan untuk meningkatkan kualitas Mesin Penerjemah Statistik (MPS) untuk bahasa bersumber daya rendah, yaitu bahasa Madura-Sunda. Dalam hal ini, bahasa Indonesia digunakan sebagai bahasa penghubung, di mana korpus yang tersedia masing-masing dapat digunakan untuk mendukung pasangan bahasa Madura-Sunda. Eksperimen yang dilakukan menggunakan korpus paralel bahasa Indonesia-Madura dan Indonesia-Sunda masing-masing sebesar sebesar 5K dan 6K kalimat, sedangkan korpus monolingual yang digunakan adalah bahasa Melayu, Sunda dan Indonesia masing-masing sebesar 10K, 10K, dan 100K kalimat. Penelitian ini membandingkan hasil penerapan metode Triangulation dan Transfer menggunakan bahasa Indonesia sebagai bahasa pivot. Hasil penelitian yang dilakukan menunjukkan bahwa metode Triangulation memiliki akurasi yang lebih baik, jika dibandingkan dengan metode Transfer. Dari hasil percobaan yang dilakukan, metode Triangulation rata-rata meningkatkan akurasi MPS berbasis pivot bahasa Indonesia sebesar 6,18% untuk MPS Madura-Sunda dan 7,27% untuk MPS Sunda-Madura.

Statistical machine translation for low-resource language suffers from the lack of abundant training corpora. Several methods, such as the use of a pivot language, have been proposed as a bridge to translate from one language to another. However, errors will accumulate during the extensive translation pipelines. In this paper, we propose an approach to low-resource language translation by exploiting the pronunciation correlations between languages. We find that the pronunciation features can improve both Chinese-Vietnamese and Vietnamese-Chinese translation qualities. Experimental results show that our proposed model yields effective improvements, and the translation performance (bilingual evaluation understudy score) is improved by a maximum value of 1.03.

Multilinguality is of particular interest for digital libraries in Cultural Heritage (CH), where the language of the data may not match users’ languages. However, multilingual access is rarely implemented beyond the use of multilingual interfaces. We have run an experiment using the Europeana CH digital library as a use case. We evaluate the effectiveness of a multilingual information retrieval strategy using machine translations to English as pivot language. We conducted an indirect evaluation that should be considered preliminary. Yet, together with a manual analysis of the query translations, it already shows (or confirms) some of the benefits and challenges of deploying such systems in CH.

Pivot translations are very often used in film festivals, but have been granted little consideration from an academic viewpoint. This article analyses the role of pivot languages in audiovisual translation within the framework of Asian film festivals held in Catalonia. There are three aims of this paper: (i) to examine to what extent pivot translations are part of the translation process in films screened in such festivals, (ii) to determine the justifications for their use, and (iii) to analyse the effects of their use from a qualitative perspective. In order to do so, the answers from a questionnaire distributed among the most relevant agents in Asian film festivals in Catalonia will be analysed. Additionally, the Chinese film Old Stone by Johnny Ma that has been translated into and subtitled in Catalan through English as its pivot language, will be presented as a case study. Lay abstract The use of a third language or pivot translation is widespread in film festivals, although very few studies focus on this practice, which usually remains unnoticed by the average spectator. This article seeks to examine just how common this phenomenon is in film festivals and to analyse its impact with a case study, taking the Chinese film Old Stone by Johnny Ma and its translation into Catalan as an example. More precisely, the article aims to answer two questions regarding the use of pivot languages in audiovisual translation. Firstly, to what extent and for what exact purpose are pivot translations currently being used in Asian film festivals in Catalonia? Secondly, how does using a pivot language, in our case English, affect quality?

This paper reports on the efforts of twelve national teams in building the International Comparable Corpus (ICC; https://korpus.cz/icc) that will contain highly comparable datasets of spoken, written and electronic registers. The languages currently covered are Czech, Finnish, French, German, Irish, Italian, Norwegian, Polish, Slovak, Swedish and, more recently, Chinese, as well as English, which is considered to be the pivot language. The goal of the project is to provide much-needed data for contrastive corpus-based linguistics. The ICC corpus is committed to the idea of re-using existing multilingual resources as much as possible and the design is modelled, with various adjustments, on the International Corpus of English (ICE). As such, ICC will contain approximately the same balance of forty percent of written language and 60 percent of spoken language distributed across 27 different text types and contexts. A number of issues encountered by the project teams are discussed, ranging from copyright and data sustainability to technical advances in data distribution.

Transfer learning brings improvement to machine translation by using a resource-rich language pair to pretrain the model and then adapting it to the desired language pair. However, to date, there have been few attempts to tackle machine transliteration with transfer learning. In this article, we propose a method of using source–pivot and pivot–target datasets to improve source–target machine transliteration. Our approach first bridges the source–pivot and pivot–target datasets by reducing the distance between source and pivot embeddings. Then, our model learns to translate from the pivot language to the target language. Finally, the source–target dataset is used to fine tune the model. Our experiments show that our method is superior to the transfer learning method. When implemented with a state-of-the-art source–target translation model from NEWS’18, our transfer learning method can improve the accuracy by 1.1%.

Exploiting a common language as an auxiliary for better translation has a long tradition in machine translation and lets supervised learning-based machine translation enjoy the enhancement delivered by the well-used pivot language in the absence of a source language to target language parallel corpus. The rise of unsupervised neural machine translation (UNMT) almost completely relieves the parallel corpus curse, though UNMT is still subject to unsatisfactory performance due to the vagueness of the clues available for its core back-translation training. Further enriching the idea of pivot translation by extending the use of parallel corpora beyond the source-target paradigm, we propose a new reference language-based framework for UNMT, RUNMT, in which the reference language only shares a parallel corpus with the source, but this corpus still indicates a signal clear enough to help the reconstruction training of UNMT through a proposed reference agreement mechanism. Experimental results show that our methods improve the quality of UNMT over that of a strong baseline that uses only one auxiliary language, demonstrating the usefulness of the proposed reference language-based UNMT and establishing a good start for the community., EMNLP 2020, ACL Findings

As a typical low-resource language pair, besides severely limited by the scale of parallel corpus, Chinese-Lao language pair also has considerable linguistic differences, resulting in poor performance of Chinese-Lao neural machine translation (NMT) task. However, compared with the Chinese-Lao language pair, there are considerable cross-lingual similarities between Thai-Lao languages. According to these features, we propose a novel NMT approach. We first train Chinese-Thai and Thai-Lao NMT models wherein Thai is treated as pivot language. Then the transfer learning strategy is used to extract the encoder and decoder respectively from the two trained models. Finally, the encoder and decoder are combined into a new model and then fine-tuned based on a small-scale Chinese-Lao parallel corpus. We argue that the pivot language Thai can deliver more information to Lao decoder via linguistic similarity and help improve the translation quality of the proposed transfer-based approach. Experimental results demonstrate that our approach achieves 9.12 BLEU on Chinese-Lao translation task using a small parallel corpus, compared to the 7.37 BLEU of state-of-the-art Transformer baseline system using back-translation.

In this paper we present the efforts to diversify Serbian-French-English-Spanish corpus ParCoLab. ParCoLab is the project led by CLLE research unit (UMR 5263 CNRS) at the University of Toulouse, France, and the Romance Department at the University of Belgrade, Serbia. The main goal of the project is to create a freely searchable and widely applicable multilingual resource with Serbian as the pivot language. Initially, the majority of the corpus texts represented written language. Since diversity of text types contributes to the usefulness and applicability of a parallel corpus, a great deal of effort has been made to include spoken language data in the ParCoLab database. Transcripts and translations of TED talks, films and cartoons have been included so far, along with transcripts of original Serbian films. Thus, the 17.6M-word database of mainly literary texts has been extended with spoken language data and it now contains 32.9M words.

Interactive system development does not follow the same life cycle as other system developments. The essential differences lie in the iterative nature of such system development. Throughout iterations, the requirements undergo many changes due to the evolution of customer’s needs and user feedback after experiencing prototypes. The challenge is tougher in the case of critical HCI (aircraft cockpits, medical systems, nuclear power plant etc.). Indeed, critical HCI requires to be designed and built such that safety and security requirements are thoroughly fulfilled. This paper presents a novel development framework featuring formal verification and validation of critical interactive systems at modelling level. It identifies the several challenges of the interactive system development and settles down the main objectives. Furthermore, it sketches the most important traits of the proposed approach to circumvent the enumerated challenges. Hence, we discuss the importance of the pivot language Fluid, the domain theory specific to interactive systems and their modelling using Event-B.

Bilingual dictionary is a useful resource for computational linguistics and natural language processing research. Especially in enriching low-resource languages, the first important step is creating bilingual dictionary. For the closely-related ones, it has been shown that the pivot-based approach is useful for inducing bilingual lexicons from two bilingual dictionaries via the pivot language. However, if there are no available machine-readable dictionaries as input, we need to consider manual creation by human. To realize comprehensive creation of multiple bilingual dictionaries, even if we already have several existing machine-readable bilingual dictionaries, it is still difficult to determine the execution order of the pivot-based approach to reducing the total cost. We formalize the plan optimization for constructing execution order of bilingual dictionary creation by utilizing Markov Decision Process (MDP) and considering methods and their costs.

On college campuses throughout the United States, classes for decades have been offered to help Spanish speakers learn Portuguese, augmented recently by similar courses in Italian, French, and Catalan. Though there exist numerous commonalities between these innovative course offerings, the pivotal role of Spanish stands out. This essay therefore introduces the concept of a pivot language: a widely spoken second language leveraged through explicit instruction to facilitate the learning of a related third language (L3). It is suggested that in an increasingly multilingual country, this trend of taking into account students' prior linguistic knowledge points to where language teaching is headed.

Machine translation, which will be used widely in human–computer interaction services to Internet of Things (IoT), is a key technology in artificial intelligence field. This paper presents a minimum Bayes-risk (MBR) phrase table pruning method for pivot-based statistical machine translation (SMT). The SMT system requires a great amount of bilingual data to build a high-performance translation model. For some language pairs, such as Chinese–English, massive bilingual data are available on the web. However, for most language pairs, large-scale bilingual data are hard to obtain. Pivot-based SMT is proposed to solve the data scarcity problem: it introduces a pivot language to bridge the source language and the target language. Therefore, a source-target translation model based on well-trained source-pivot and pivot-target translation models can be derived with the pivot-based approach. However, due to the ambiguities of the pivot language, source and target phrases with different meanings may be wrongly matched. Consequently, the derived source-target phrase table may contain incorrect phrase pairs. To alleviate this problem, we apply the MBR method to prune the phrase table. The MBR pruning method removes the phrase pairs with the lowest risk from the phrase table. Experimental results on Europarl data show that the proposed method can both reduce the size of phrase tables and improve the performance of translations. This study also gives a useful reference to many IoT research field and smart web services.

Zero-shot neural machine translation (NMT) is a framework that uses source-pivot and target-pivot parallel data to train a source-target NMT system. An extension to zero-shot NMT is zero-resource NMT, which generates pseudo-parallel corpora using a zero-shot system and further trains the zero-shot system on that data. In this paper, we expand on zero-resource NMT by incorporating monolingual data in the pivot language into training; since the pivot language is usually the highest-resource language of the three, we expect monolingual pivot-language data to be most abundant. We propose methods for generating pseudo-parallel corpora using pivot-language monolingual data and for leveraging the pseudo-parallel corpora to improve the zero-shot NMT system. We evaluate these methods for a high-resource language pair (German-Russian) using English as the pivot. We show that our proposed methods yield consistent improvements over strong zero-shot and zero-resource baselines and even catch up to pivot-based models in BLEU (while not requiring the two-pass inference that pivot models require).

International audience; In this article, we present the conceptual foundations and implementation principles of model synchronization, a formal framework for the management of heterogeneous models. The proposed approach relies on S2ML (System Structure Modeling Language) as a pivot language. We show, by means of a case study, that model synchronization can be used to ensure the consistency between system architecture models designed with Capella and safety models written in AltaRica 3.0.

Generating image descriptions in different languages is essential to satisfy users worldwide. However, it is prohibitively expensive to collect large-scale paired image-caption dataset for every target language which is critical for training descent image captioning models. Previous works tackle the unpaired cross-lingual image captioning problem through a pivot language, which is with the help of paired image-caption data in the pivot language and pivot-to-target machine translation models. However, such language-pivoted approach suffers from inaccuracy brought by the pivot-to-target translation, including disfluency and visual irrelevancy errors. In this paper, we propose to generate cross-lingual image captions with self-supervised rewards in the reinforcement learning framework to alleviate these two types of errors. We employ self-supervision from mono-lingual corpus in the target language to provide fluency reward, and propose a multi-level visual semantic matching model to provide both sentence-level and concept-level visual relevancy rewards. We conduct extensive experiments for unpaired cross-lingual image captioning in both English and Chinese respectively on two widely used image caption corpora. The proposed approach achieves significant performance improvement over state-of-the-art methods., Accepted by ACMMM 2019

Pivoting through a popular language with more parallel corpora available (e.g. English and Chinese) is a common approach to build machine translation (MT) systems for low-resource languages. For example, to build a Russian-to-Spanish MT system, we could build one system using the Russian–Spanish corpus directly. We could also build two systems, Russian-to-English and English-to-Spanish, as the resources of the two language pairs are much larger than the Russian–Spanish pair, and use them cascadingly to translate texts in Russian into Spanish by pivoting through English. There are, however, some confusing results on the Pivot MT approach in the literature. In this paper, we reviewed the performance of Pivot MT with the United Nations Parallel Corpus v1.0 (UN6Way) using both English and Chinese as pivot languages. We also report our system performance on the CWMT 2018 Pivot MT shared task, where Japanese patent sentences are translated into English using Chinese as the pivot language.

This paper presents the participation of DBMS-KU Interpolation system in WMT19 shared task, namely, Kazakh-English language pair. We examine the use of interpolation method using a different language model order. Our Interpolation system combines a direct translation with Russian as a pivot language. We use 3-gram and 5-gram language model orders to perform the language translation in this work. To reduce noise in the pivot translation process, we prune the phrase table of source-pivot and pivot-target. Our experimental results show that our Interpolation system outperforms the Baseline in terms of BLEU-cased score by +0.5 and +0.1 points in Kazakh-English and English-Kazakh, respectively. In particular, using the 5-gram language model order in our system could obtain better BLEU-cased score than utilizing the 3-gram one. Interestingly, we found that by employing the Interpolation system could reduce the perplexity score of English-Kazakh when using 3-gram language model order.

We propose a novel model architecture and training algorithm to learn bilingual sentence embeddings from a combination of parallel and monolingual data. Our method connects autoencoding and neural machine translation to force the source and target sentence embeddings to share the same space without the help of a pivot language or an additional transformation. We train a multilayer perceptron on top of the sentence embeddings to extract good bilingual sentence pairs from nonparallel or noisy parallel data. Our approach shows promising performance on sentence alignment recovery and the WMT 2018 parallel corpus filtering tasks with only a single model., Comment: ACL 2019 Repl4NLP camera-ready

We present effective pre-training strategies for neural machine translation (NMT) using parallel corpora involving a pivot language, i.e., source-pivot and pivot-target, leading to a significant improvement in source-target translation. We propose three methods to increase the relation among source, pivot, and target languages in the pre-training: 1) step-wise training of a single model for different language pairs, 2) additional adapter component to smoothly connect pre-trained encoder and decoder, and 3) cross-lingual encoder training via autoencoding of the pivot language. Our methods greatly outperform multilingual models up to +2.6% BLEU in WMT 2019 French-German and German-Czech tasks. We show that our improvements are valid also in zero-shot/zero-resource scenarios., EMNLP 2019 camera-ready

Článek se zabývá vícejazyčným rozpoznávání dialogových aktů. Navržené metody jsou založeny na hlubokých neuronových sítích a používají word2vec slovní vektory pro reprezentaci slov. Prezentovány jsou dvě metody. První z nich používá jeden obecný model trénovaný na všech dostupných jazycích, zatímco druhý se natrénuje pouze na jednom jazyku a pomocí lineární transformace dochází k projekci prostorů na zvolený jazyk. Jako klasifikátor jsou použity populární konvoluční neuronové sítě a LSTM. Dle našeho mínění jde o jednu z prvních metod pro vícejazyčnou klasifikaci dialogových aktů pomocí neuronových sítí. Modely jsou testovány experimentálně na dvoujazyčném korpusu Verbmobil This paper deals with multi-lingual dialogue act (DA) recognition. The proposed approaches are based on deep neural networks and use word2vec embeddings for word representation. Two multi-lingual models are proposed for this task. The first approach uses one general model trained on the embeddings from all available languages. The second method trains the model on a single pivot language and a linear transformation method is used to project other languages onto the pivot language. The popular convolutional neural network and LSTM architectures with different set-ups are used as classifiers. To the best of our knowledge this is the first attempt at multi-lingual DA recognition using neural networks. The multi-lingual models are validated experimentally on two languages from the Verbmobil corpus.

Machine Translation (MT) has been one of the classic AI tasks from the early days of the field. Portuguese and Chinese are languages with a very large number of native speakers, though this does not carry through to the amount of literature on their processing, or to the amount of resources available to be used, in particular when compared with English. In this paper, we address the feasibility of creating a MT system for Portuguese-Chinese, using only freely available resources, by experimenting with various approaches to pairing source and target parallel data during training. These approaches are (i) using a model for each source-target language pair, (ii) using an intermediate pivot language, and (iii) using a single model that can translate from any language seen in the source side to any language seen on the target side. We find approaches whose performance is higher than that of the strong baseline consisting of an MT service provided by an IT industry giant for the pair Portuguese-Chinese.

In the Industry 4.0 context, the high demand for the integration of new emerging IT technologies into production systems requires their designers to find effective ways to manage the impact of these changes during the design phase, while meeting very tight time constraints. Based on the model transformation concept at the SysML (System Modeling Language) diagrams scale, we propose to define a pivot language, through the sequence diagram, to automatically ensure the consistency between the different diagrams used throughout a MBSE (Model-Based System Engineering) design methodology, and then to guarantee the consistency of the relative design artefacts. The transformation rules have been defined between the underlying elements of SysML and could be modified according to the methodology. Finally, some rules were implemented in the PTC Modeler tool to validate the approach.

This article innovatively addresses machine translation from Chinese to Catalan using neural pivot strategies trained without any direct parallel data. The Catalan language is very similar to Spanish from a linguistic point of view, which motivates the use of Spanish as pivot language. Regarding neural architecture, we are using the latest state-of-the-art, which is the Transformer model, only based on attention mechanisms. Additionally, this work provides new resources to the community, which consists of a human-developed gold standard of 4,000 sentences between Catalan and Chinese and all the others United Nations official languages (Arabic, English, French, Russian, and Spanish). Results show that the standard pseudo-corpus or synthetic pivot approach performs better than cascade.

Term translation of Chinese historical classics is very difficult and time-consuming work, and using term alignment methods to extract term translation pairs is of great help for historical term translation. However, the limited bilingual corpora resources of historical classics and special morphology of the ancient Chinese result in poor performance of term alignment. To this end, this paper proposes a historical term alignment method using modern Chinese as a pivot language. The method first identifies English terms by rules, then aligns them from English to modern Chinese and then from modern Chinese to ancient Chinese. The use of English-modern Chinese corpus and modern-ancient Chinese corpus instead of English-ancient Chinese corpus solves the shortage problem of the parallel corpus. Moreover, using modern Chinese as a pivot language effectively reduces the alignment errors caused by the abbreviations and the interchangeable characters of ancient Chinese. In the term alignment experiment on Shiji, our method outperformed the direct alignment method significantly, which proves the validity of our method.

In view of the most difficult problem of terminology translation in the translation of historical books, this paper extracts term translation pairs from bilingual parallel corpus as a reference to improve the quality and efficiency of the translation of historical books. However, the small size of the corpus of historical books leads to poor word alignment, which brings a large number of errors to term alignment. At the same time, the special grammar of ancient Chinese will also affect term alignment. In order to solve the above two problems, this paper proposes a method of term alignment of historical books based on pivot language and position correspondence. This method introduces modern Chinese as the pivot language, makes use of the position correspondence between modern Chinese and English terms in sentences, and uses the sliding window method to generate candidate term pairs, which avoids the influence of word alignment errors on term alignment. In addition, if we first align to modern Chinese, the ancient Chinese terminology will complete the omitted characters in modern Chinese and replace the false characters, so as to avoid the influence of the special usage in ancient Chinese on the term alignment. The experimental results in Historical Records show that the accuracy of this method is slightly better than that of the method based on word alignment, which verifies the effectiveness of this method without word alignment.

In the age of Web 2.0, community user contributed questions and answers provide an important alternative for knowledge acquisition through web search. Question retrieval in current community-based question answering (CQA) services do not, in general, work well for long and complex queries, such as the questions. The main reasons are the verboseness in natural language queries and the word mismatch between the queries and the candidate questions in the CQA archive during retrieval. To address these two problems, existing solutions try to refine the search queries by distinguishing the key concepts in the queries and expanding the queries with relevant content. However, using the existing query refinement approaches can only identify the key and non-key concepts, while the differences between the key concepts are overlooked. Moreover, the existing query expansion approaches, not only overlook the weights of key concepts in the queries, but also fail to consider concept level expansion for them. In this paper, we explore a key concept identification approach for query refinement and a pivot language translation based approach to explore key concept paraphrasing. We further propose a new question retrieval model which can seamlessly integrate the key concepts and their paraphrases. The experimental results demonstrate that the integrated retrieval model significantly outperforms the state-of-the-art models in question retrieval.

The paper describes a rule-based machine translation (MT) system from North to South Saami. The system is designed for a workflow where North Saami functions as pivot language in translation from Norwegian or Swedish. We envisage manual translation from Norwegian or Swedish to North Saami, and thereafter MT to South Saami. The system was aimed at a single domain, that of texts for use in school administration. We evaluated the system in terms of the quality of translations for postediting. Two out of three of the Norwegian to South Saami professional translators found the output of the system to be useful. The evaluation shows that it is possible to make a functioning rule-based system with a small transfer lexicon and a small number of rules and achieve results that are useful for a restricted domain, even if there are substantial differences b etween the languages.

We investigate the effectiveness of neural machine translation in a low-resource setting, where particular attention is given to preserving the rhetorical style of a given persona. In particular, we attempt to complete Cicero’s translation of Timaeus, a text originally written by Plato in Attic Greek. Due to the scarcity of parallel Attic Greek and Classical Latin text, we use English as a pivot language. Finally, we explore how additional texts, which may help improve the accuracy of the model, affect the preservation of a particular persona (i.e. Socrates) through translation into the pivot language and subsequently the target language.

International audience; Cyber physical systems are specified in a hybrid form, with discrete and continuous parts. Simulating such systems requires precise data and synchronization of continuous changes and discrete transitions. However, in the first design steps, missing information forbids numerical simulation. We present here a new qualitative language for qualitative simulation of cyber physical systems, which consists in computing the relationships between the system variables. This language is implemented in the Diversity symbolic execution engine to build the traces of the system. We apply this approach to the analysis of SysML models, using an M2M transformation from SysML to a pivot language, an M2T transformation from this language to Diversity and an analysis of the traces obtained by Diversity to build the qualitative behaviors of the system.

International audience; Hybrid systems are specified in a heterogeneous form, with discrete and continuous parts. Simulating such systems requires precise data and synchronization of continuous changes and discrete transitions. However, in the early design stages, missing information forbids numerical simulation. We present here a symbolic execution model for the qualitative simulation of hybrid systems, which consists in computing only qualities of the behavior. This model is implemented in the Diversity symbolic execution engine to build the qualitative behaviors of the system. We apply this approach to the analysis of SysML models, using an M2M transformation from SysML to a pivot language and an M2T transformation from this language to Diversity.

Image captioning is a multimodal task involving computer vision and natural language processing, where the goal is to learn a mapping from the image to its natural language description. In general, the mapping function is learned from a training set of image-caption pairs. However, for some language, large scale image-caption paired corpus might not be available. We present an approach to this unpaired image captioning problem by language pivoting. Our method can effectively capture the characteristics of an image captioner from the pivot language (Chinese) and align it to the target language (English) using another pivot-target (Chinese-English) sentence parallel corpus. We evaluate our method on two image-to-English benchmark datasets: MSCOCO and Flickr30K. Quantitative comparisons against several baseline approaches demonstrate the effectiveness of our method.

Bilingual lexicon is a useful language resource, but such data rarely available for lower-density language pairs, especially for those that are closely related. The lack or absence of parallel and comparable corpora makes bilingual lexicon extraction becomes a difficult task. Using a third language to link two other languages is a well-known solution in low-resource situation, which usually requires only two input bilingual lexicons to automatically induce the new one. This approach, however, is weak in measuring semantic distance between bilingual word pairs because it has never been demonstrated to utilize the complete structures of the input bilingual lexicons as dropped meanings negatively influence the result. This research discuss a constraint approach to pivot-based lexicon induction in case the target language pair are closely related. We create constraints from language similarity and model the structures of the input dictionaries as an optimization problem whose solution produces optimally correct target bilingual lexicon. In addition, we enable created bilingual lexicons of low-resource languages accessible through service grid federation.