Machine Translation (MT) Quality Estimation (QE) is the task of determining the quality of an automatic translation given its source sentence and without recourse to reference translations. The automatic estimation of machine translation output quality is a hard task in which the selection of the appropriate algorithm and the most predictive features over reasonably sized training sets plays a crucial role. When moving from controlled lab evaluations to real-life scenarios the task becomes even harder. For current MT quality estimation systems, additional complexity comes from the difficulty to model user and domain changes. The instability of the systems with respect to data coming from different distributions, in fact, calls for adaptive solutions that react to new operating conditions. To tackle this issue we developed AQET, an online framework for adaptive QE that targets reactivity and robustness to user and domain changes.
AQET (Adaptive Quality Estimation Tool) is an open-source package for performing Quality Estimation for Machine Translation able to continuously learn from post-edited sentences. AQET has been developed to support professional translators during their daily work and it is suitable for being embedded in a Computer-Assisted Translation tool. The currentl version (v1.0) supports two online machine learning algorithms: Online Support Vector Regression (Online SVR) and Passive-Aggressive. More information on the software are available here. This work has been partially supported by the EC- funded project MateCat (ICT-2011.4.2-287688).
How To obtain AQET
AQET is open-source distributed under the GPL v.3 license.
AQET can be downloaded here.
Reference paper
whenever making reference to this software, please cite the following paper:
M. Turchi, A. Anastasopoulos, J. G. Camargo de Souza and M. Negri “Adaptive Quality Estimation for Machine Translation“, In Proceedings of the the 52nd Annual Meeting of the Association for Computational Linguistics, Baltimore (ACL), Usa 22-27 June 2014.
@InProceedings{turchi-EtAl:2014:P14-1,
author = {Turchi, Marco and Anastasopoulos, Antonios and C. de Souza, Jos\'{e} G. and Negri, Matteo},
title = {Adaptive Quality Estimation for Machine Translation},
booktitle = {Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
month = {June},
year = {2014},
address = {Baltimore, Maryland},
publisher = {Association for Computational Linguistics},
pages = {710–720},
url = {http://www.aclweb.org/anthology/P/P14/P14-1067}
}
External Software
AQET takes advantage of third-party open-source software:
- QuEst: an open source tool for translation quality estimation
- Online SVR: C++ implementation of Online Support Vector Regression algorithm
- sofia-ml: suite of fast incremental algorithms for machine learning
- tercpp: C++ implementation of TER metric
Contact us: for more infomation please contact: turchi[at]fbk.eu or anastasopoulos.ant[at]gmail.com or negri[at]fbk.eu
