No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Korean-Centered Cross-Lingual Parallel Sentence Corpus Construction Experiment
ResearchGate has not been able to resolve any citations for this publication.
We introduce an architecture to learn joint multilingual sentence representations for 93 languages, belonging to more than 30 different families and written in 28 different scripts. Our system uses a single BiLSTM encoder with a shared byte-pair encoding vocabulary for all languages, which is coupled with an auxiliary decoder and trained on publicly available parallel corpora. This enables us to learn a classifier on top of the resulting embeddings using English annotated data only, and transfer it to any of the 93 languages without any modification. Our experiments in cross-lingual natural language inference (XNLI data set), cross-lingual document classification (MLDoc data set), and parallel corpus mining (BUCC data set) show the effectiveness of our approach. We also introduce a new test set of aligned sentences in 112 languages, and show that our sentence embeddings obtain strong results in multilingual similarity search even for low- resource
languages. Our implementation, the pre-trained encoder, and the multilingual test set are available at https://github.com/facebookresearch/LASER .
In this paper, we present an approach to learn multilingual sentence embeddings using a bi-directional dual-encoder with additive margin softmax. The embeddings are able to achieve state-of-the-art results on the United Nations (UN) parallel corpus retrieval task. In all the languages tested, the system achieves P@1 of 86% or higher. We use pairs retrieved by our approach to train NMT models that achieve similar performance to models trained on gold pairs. We explore simple document-level embeddings constructed by averaging our sentence embeddings. On the UN document-level retrieval task, document embeddings achieve around 97% on P@1 for all experimented language pairs. Lastly, we evaluate the proposed model on the BUCC mining task. The learned embeddings with raw cosine similarity scores achieve competitive results compared to current state-of-the-art models, and with a second-stage scorer we achieve a new state-of-the-art level on this task.
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 1,100 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.
Researchers in both machine translation (e.g., Brown ., 1990) and bilingual lexicography (e.g., Klavans and Tzoukermann, 1990) have recently become interested in studying parallel texts, texts such as the Canadian Hansards (parliamentary proceedings) which are available in multiple languages (French and English). This paper describes a method for aligning sentences in these parallel texts, based on a simple statistical model of character lengths. The method was developed and tested on a small trilingual sample of Swiss economic reports. A much larger sample of 90 million words of Canadian Hansards has been aligned and donated to the ACL/DCI.
The dominant sequence transduction models are based on complex recurrent or convolutional neural networks in an encoder-decoder configuration. The best performing models also connect the encoder and decoder through an attention mechanism. We propose a new simple network architecture, the Transformer, based solely on attention mechanisms, dispensing with recurrence and convolutions entirely. Experiments on two machine translation tasks show these models to be superior in quality while being more parallelizable and requiring significantly less time to train. Our model achieves 28.4 BLEU on the WMT 2014 English-to-German translation task, improving over the existing best results, including ensembles by over 2 BLEU. On the WMT 2014 English-to-French translation task, our model establishes a new single-model state-of-the-art BLEU score of 41.0 after training for 3.5 days on eight GPUs, a small fraction of the training costs of the best models from the literature. We show that the Transformer generalizes well to other tasks by applying it successfully to English constituency parsing both with large and limited training data.
We address the problem of unsupervised and language-pair independent alignment of symmetrical and asymmetrical parallel corpora. Asymmetrical parallel corpora contain a large proportion of 1-to-0/0-to-1 and 1-to-many/many-to-1 sentence corre- spondences. We have developed a novel approach which is fast and allows us to achieve high accuracy in terms of F1 for the alignment of both asymmetrical and symmetrical parallel corpora. The source code of our aligner and the test sets are freely available.
Cross-lingual Language Model Pretraining
- A Conneau
- G Lample
Korean Language Resources for Everyone
- Park
Korean Language Resources for Everyone
- J Park
- J.-P Hong
- J.-W Cha