Review of Cyril Goutte, Nicola Cancedda, Marc Dymetman, and George Foster (eds): Learning machine translation

摘要

As the field of statistical machine translation matures, there is not only a need to develop better linguistic models but also better parameter estimation methods. Not only must we develop models that match more closely the linguistic processes to explain the transformation of a text in one language into a meaning-equivalent text in another. There is also a need to refine the core methodology of obtaining parameter values for such models.