Statistical Machine Translation and Comparable Corpora
Krzysztof Wołk · 2019
This chapter provides a brief overview of statistical machine translation, and discusses textual components and corpora. It describes the Moses tool environment for statistical machine translation (SMT), examines major aspects of SMT processing, and explores the evaluation of SMT output. SMT is a method of translation in which a translation is created on the basis of statistical models, which are obtained through an analysis of parallel corpus of a language pair. The multilingual nature of the world makes translation a crucial requirement today. Parallel dictionaries constructed by humans are a widely-available resource, but they are limited and do not provide enough coverage for good quality translation purposes, due to out-of-vocabulary words and neologisms. Moses is a tool environment for statistical machine translation that enables users to train translation models for any two languages. The best way to obtain great performance from an SMT system is by using a decent translation phrase table.