Exploiting Similarities among Languages for Machine Translation
This method addresses the bottleneck of manual dictionary creation for machine translation systems, applicable to any language pair, though it is incremental as it builds on existing statistical translation frameworks.
The paper tackled the problem of automating dictionary and phrase table generation for machine translation by learning language structures from monolingual data and mapping between languages from bilingual data, achieving almost 90% precision@5 for English-Spanish word translation.
Dictionaries and phrase tables are the basis of modern statistical machine translation systems. This paper develops a method that can automate the process of generating and extending dictionaries and phrase tables. Our method can translate missing word and phrase entries by learning language structures based on large monolingual data and mapping between languages from small bilingual data. It uses distributed representation of words and learns a linear mapping between vector spaces of languages. Despite its simplicity, our method is surprisingly effective: we can achieve almost 90% precision@5 for translation of words between English and Spanish. This method makes little assumption about the languages, so it can be used to extend and refine dictionaries and translation tables for any language pairs.