User‐Centered Evaluation of Lexicons Extracted from Comparable Corpora
Estelle Maryline Delpech · 2014
This chapter proposes an applicative evaluation protocol of the bilingual lexicons extracted from comparable corpora, intended for specialized translation. It focuses on evaluation methodology used both in MT and in human translation. The chapter exposes the assessment protocol that describes its implementation as well as the results obtained. It looks at the results of experiment and suggests new research paths to explore. The MT evaluation has two purposes. First, it analyzes, during the development of an MT system, the impact of a system modification on the quality of the translations. Second, the evaluation enables to compare the systems between them, usually during a broad evaluation campaign. In translation, the question of evaluation is referred as translation quality assessment (TQA).