Google Converts Language Translation into a Problem of Vector Space Mathematics

In 2013 Tomas Mikolov, Quoc V. Le, and Ilya Sutskever at Google in Mountain View developed a technique that automatically generated dictionaries and phrase tables that convert one language into another. Instead of relying on versions of the same document in different languages the new method used data mining techniques to model the structure of a single language, and then compared it to the structure of another language.

ABSTRACT:
"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 pair"

 

Mikolov et al, "Exploiting Simularities among Languages for Machine Translation" (2013) http://arxiv.org/pdf/1309.4168v1.pdf, accessed 01-14-2015).

Tomas Mikolov, Kai Chen, Greg Corrado, Jeffrey Dean, "Efficient Estimation of Word Representations in Vector Space", (September7,2013)