Unsupervised Multilingual Sentence Embeddings for Parallel Corpus Mining
This addresses the challenge of low-resource languages lacking parallel data for multilingual NLP applications, though it is incremental as it builds on existing unsupervised translation techniques.
The authors tackled the problem of deriving multilingual sentence embeddings without parallel data by proposing an unsupervised method that uses monolingual data and synthetic parallel corpora, resulting in improvements of up to 22 F1 points on parallel corpus mining tasks.
Existing models of multilingual sentence embeddings require large parallel data resources which are not available for low-resource languages. We propose a novel unsupervised method to derive multilingual sentence embeddings relying only on monolingual data. We first produce a synthetic parallel corpus using unsupervised machine translation, and use it to fine-tune a pretrained cross-lingual masked language model (XLM) to derive the multilingual sentence representations. The quality of the representations is evaluated on two parallel corpus mining tasks with improvements of up to 22 F1 points over vanilla XLM. In addition, we observe that a single synthetic bilingual corpus is able to improve results for other language pairs.