Introduction

Inguna Skadina, Robert J. Gaizauskas, Andrejs Vasiljevs, Monica Lestari Paramita

Abstract: This book addresses the full set of questions that arise when attempting to exploit comparable corpora to overcome the bottleneck of insufficient parallel corpora that affects any data-driven machine translation approach, particularly in relation to under-resourced languages and narrow domains. It describes methods and tools for identifying and assessing comparability, for gathering comparable corpora from the Web, for extracting translation equivalents from within comparable texts and discusses the evaluation of this pipeline of methods and tools by incorporating their outputs into a machine translation system and assessing its performance in real application settings.