Unsupervised Hyperalignment for Multilingual Word Embeddings
This addresses the challenge of multilingual NLP tasks by enhancing cross-lingual word alignment, though it is incremental as it builds on prior unsupervised methods for two languages.
The paper tackles the problem of aligning word embeddings across multiple languages to a common space without supervision, proposing a novel formulation that ensures composable mappings to improve indirect word translation quality, showing consistent improvements in evaluations across eleven languages.
We consider the problem of aligning continuous word representations, learned in multiple languages, to a common space. It was recently shown that, in the case of two languages, it is possible to learn such a mapping without supervision. This paper extends this line of work to the problem of aligning multiple languages to a common space. A solution is to independently map all languages to a pivot language. Unfortunately, this degrades the quality of indirect word translation. We thus propose a novel formulation that ensures composable mappings, leading to better alignments. We evaluate our method by jointly aligning word vectors in eleven languages, showing consistent improvement with indirect mappings while maintaining competitive performance on direct word translation.