Word Confidence Estimation and its Integration in Sentence Quality Estimation for Machine Translation

Ngoc-Quang Luong (1) , Laurent Besacier (1) , Benjamin Lecouteux (1)

1 GETALP - Groupe d’Étude en Traduction Automatique/Traitement Automatisé des Langues et de la Parole

This paper proposes some ideas to build an effective estima-tor, which predicts the quality of words in a Machine Translation (MT) output. We integrate a number of features of various types (system-based, lexical, syntactic and semantic) into the conventional feature set, for our baseline classifier training. Once having experiments with all features , we deploy a " Feature Selection " strategy to filter the best performing ones. Then, a method that combines multiple " weak " classifiers to build a strong " composite " classifier by taking advantage of their com-plementarity allows us achieve a better performance in term of F score. Finally, we exploit word confidence scores for improving the estimation system at sentence level.

Mots clés: Confidence Measure; Confidence Es- timation; Conditional Random Fields; Boosting; Machine Translation

Domaines: Intelligence artificielle [cs.AI]

Ngoc-Quang Luong, Laurent Besacier, Benjamin Lecouteux. Word Confidence Estimation and its Integration in Sentence Quality Estimation for Machine Translation. Proceedings of the fifth international conference on knowledge and systems engineering (KSE), 2013, Hanoi, Vietnam. pp.x-x. ⟨hal-00953774⟩