LIG System for Word Level QE task at WMT14

By Ngoc-Quang Luong, Laurent Besacier, Benjamin Lecouteux. 2014

29 citations recorded by the source corpus.

This paper describes our Word-level QE system for WMT 2014 shared task on Spanish-English pair. Compared to WMT 2013, this year's task is different due to the lack of SMT setting information and additional resources. We report how we overcome this challenge to retain most of the important features which performed well last year in our system. Novel features related to the availability of multiple systems output (new point of this year) are also proposed and experimented along with baseline set. The system is optimized by several ways: tuning the classification threshold, combining with WMT 2013 data, and refining using Feature Selection strategy on our development set, before dealing with the test set for submission.

Subfields: Artificial Intelligence, Natural Language Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Natural Language Processing

Citations in the corpus

Findings of the 2016 Conference on Machine Translation (2016, methodology): “Our baseline set of features is loosely based on the one described in (Luong et al., 2014).”
Findings of the 2015 Workshop on Statistical Machine Translation (2015, positive): “Our baseline set of features is loosely based on the one described in (Luong et al., 2014).”
Findings of the 2014 Workshop on Statistical Machine Translation (2014, background): “, 2014) LIG Laboratoire d’Informatique Grenoble, France (Luong et al., 2014) LIMSI Laboratoire d’Informatique pour la Mécanique et les Sciences de l’Ingénieur, France (Wisniewski et al.”
Findings of the 2017 Conference on Machine Translation (WMT17) (2017, methodology): “22 of them were taken from the feature set described in (Luong et al., 2014), and had also been used as a baseline feature set at WMT16: • Word count in the source and target sentences, and source and target token count ratio.”
MARMOT: A Toolkit for Translation Quality Estimation at the Word Level (2016, methodology): “In (Luong et al., 2014) the feature set contains source and target contexts of the word, source and target POS-tags, LM scores, syntactic and semantic information.”
Phrase-level Quality Estimation for Machine Translation (2015, methodology): “– LIG [17] — system with 25 black-box features and was trained with CRF.”
Pushing the Limits of Translation Quality Estimation (2017, background): “Past approaches to this problem include linear classiﬁers with handcrafted features (Uefﬁng and Ney, 2007; Bic¸ici, 2013; Shah et al., 2013; Luong et al., 2014), sometimes subject to feature selection (Avramidis, 2012; Beck et al., 2013), recurrent neural networks (de Souza et al., 2014; Kim and…”
LORIA System for the WMT15 Quality Estimation Shared Task (2015, methodology): “(Luong et al., 2014; Wisniewski et al., 2014) works are applied to word-level Quality Estimation while we deal with sentence-level Quality Estimation.”
Data enhancement and selection strategies for the word-level Quality Estimation (2015, methodology): “We used a subset of features described by Luong et al. (2014), mainly the features that were listed as the most informative.”
Metrics for Evaluation of Word-level Machine Translation Quality Estimation (2016, background): “…estimation (i.e. estimation of how conﬁdent a particular MT system is on a word or a phrase (Gan-drabur and Foster, 2003)) it evolved to system-independent QE and is performed at the word level (Luong et al., 2014), sentence level (Shah et al., 2013) and document level (Scarton et al., 2015).”
Strategy-Based Technology for Estimating MT Quality (2015, methodology): “The features used in this paper were from portion of features provided by organizer and portion of (Luong et al., 2014) features.”
Multi-level Translation Quality Prediction with QuEst++ (2015, methodology): “Word-level QE (Blatz et al., 2004; Luong et al., 2014) is seemingly a more challenging task where a quality label is to be produced for each target word.”