Η Αυτοματοποιημένη και μη-αυτοματοποιημένη αξιολόγηση συστήματος Στατιστικής Μηχανικής Μετάφρασης για το γλωσσικό ζεύγος Ελληνικά - Ιταλικά

Abstract

Machine Translation (MT) evaluation is a hard task considering the difficulties that raise from the translation process itself. In this paper we present the results of the evaluation of a Statistical Machine Translation (SMT) system in which the Moses decoder was trained for the language pair Greek-Italian. The evaluation task was both automatic and non–automatic (human). For the automatic evaluation, the metrics BLEU, NIST were used, while for the human evaluation, the adequacy and the fluency of the translated texts was evaluated. A corpus of 120 individual sentences were evaluated, (e.g. EU texts, scientific technical texts, subtitles, proverbs etc.), by postgraduate students of the direction of Translation, Interpretation and Communication of the Department of Italian Language and Literature of the Aristotle University of Thessaloniki. The first results show that SMT performs well when translating text of this type