Weakly supervised approaches for quality estimation - Archive ouverte HAL Access content directly
Journal Articles Machine Translation Year : 2013

Weakly supervised approaches for quality estimation

Abstract

Currently, quality estimation (QE) is mostly addressed using supervised learning approaches. In this paper we show that unsupervised and weakly supervised approaches (using a small training set) perform almost as well as supervised ones, for a significantly lower cost. More generally, we study the various possible definitions, parameters, evaluation methods and approaches for QE, in order to show that there are multiple possible configurations for this task.
Fichier principal
Vignette du fichier
mt.pdf (583.63 Ko) Télécharger le fichier
Origin : Files produced by the author(s)
Loading...

Dates and versions

hal-00860680 , version 1 (12-09-2014)

Identifiers

Cite

Erwan Moreau, Carl Vogel. Weakly supervised approaches for quality estimation. Machine Translation, 2013, 27 (3), pp 257-280. ⟨10.1007/s10590-013-9142-8⟩. ⟨hal-00860680⟩
75 View
229 Download

Altmetric

Share

Gmail Facebook Twitter LinkedIn More