Building a Better Bitext for Structurally Different Languages through Self-training
We propose a novel method to bootstrap the construction of parallel corpora for new pairs of structurally different languages. We do so by combining the use of a pivot language and self-training. A pivot language enables the use of existing translation models to bootstrap the alignment and a self-training procedure enables to achieve better alignment, both at the document and sentence level. We also propose several evaluation methods for the resulting alignment.
Code (0)
등록된 구현이 없습니다.
Tasks
Information RetrievalMachine TranslationSentenceTranslationSimilar Papers 제목 키워드 기반
xSIM++: An Improved Proxy to Bitext Mining Performance for Low-Resource Languages
We introduce a new proxy score for evaluating bitext mining based on similarity in a multilingual embedding space: xSIM++. In comparison to xSIM, this improved proxy leverages rule-based approaches to extend English sent…
NMTBeyond English-Centric Bitexts for Better Multilingual Language Representation Learning
In this paper, we elaborate upon recipes for building multilingual representation models that are not only competitive with existing state-of-the-art models but are also more parameter efficient, thereby promoting better…
Representation LearningBitext Mining for Low-Resource Languages via Contrastive Learning
Mining high-quality bitexts for low-resource languages is challenging. This paper shows that sentence representation of language models fine-tuned with multiple negatives ranking loss, a contrastive objective, helps retr…
Contrastive LearningSentenceBitext Mining Using Distilled Sentence Representations for Low-Resource Languages
Scaling multilingual representation learning beyond the hundred most frequent languages is challenging, in particular to cover the long tail of low-resource languages. A promising approach has been to train one-for-all m…
Cross-Lingual TransferNMTRepresentation LearningSentenceProducing Monolingual and Parallel Web Corpora at the Same Time - SpiderLing and Bitextor's Love Affair
This paper presents an approach for building large monolingual corpora and, at the same time, extracting parallel data by crawling the top-level domain of a given language of interest. For gathering linguistically releva…