Multi-level Translation Quality Prediction with QuEst++

By Lucia Specia, Gustavo Paetzold, Carolina Scarton. 2015

140 citations recorded by the source corpus.

This paper presents QUEST++ , an open source tool for quality estimation which can predict quality for texts at word, sentence and document level. It also provides pipelined processing, whereby predictions made at a lower level (e.g. for words) can be used as input to build models for predictions at a higher level (e.g. sentences). QUEST++ allows the extraction of a variety of features, and provides machine learning algorithms to build and test quality estimation models. Results on recent datasets show that QUEST++ achieves state-of-the-art performance.

Subfields: Artificial Intelligence, Machine Translation

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 Machine Translation

Citations in the corpus

SimpleNets: Quality Estimation with Resource-Light Neural Networks (2016, methodology): “The or-ganisers also provided 17 baseline feature values extracted using QuEst++ (Specia et al., 2015) for each dataset.”
MARMOT: A Toolkit for Translation Quality Estimation at the Word Level (2016, methodology): “Despite many submissions to the WMT word-level QE task, only one tool is publicly available for QE at the word level, namely QuEst++ (Specia et al., 2015), which is an extension of a previous, sentence-level version for QE at the document and word levels.”
Enhancement of Encoder and Attention Using Target Monolingual Corpora in Neural Machine Translation (2018, methodology): “The features of the SVM classiﬁer include the 17 basic features of QuEst++ (Specia et al., 2015).”
USHEF and USAAR-USHEF participation in the WMT15 QE shared task (2015, methodology): “The ﬁrst set contains document-aware features, based on QuEst features for sentence-level QE (Specia et al., 2013; Specia et al., 2015).”
SHEF-NN: Translation Quality Estimation with Neural Networks (2015, methodology): “For the SHEF-QuEst++ system, we combined all 40 features described in (Specia et al., 2015) with the source-to-target similarity feature described in Section 3.”
Bilingual Embeddings and Word Alignments for Translation Quality Estimation (2016, methodology): “The features set is extracted using QuEst++ 1 (Specia et al., 2015), an open source implementation of the baseline for quality estimation for different granularities (sentence, word, and document level QE).”
Sheffield Systems for the English-Romanian WMT Translation Task (2016, methodology): “For each translation candidate, we extend its feature set by adding 17 new features corresponding to the base-line black-box QE features 4 extracted with the QuEst++ toolkit 5 (Specia et al., 2015).”
H2@BUCC18: Parallel Sentence Extraction from Comparable Corpora Using Multilingual Sentence Embeddings (2018, methodology): “Learning features: We use the following group of features which have been used in work related to translation quality estimation for several languages (Bouamor et al., 2013b; Specia et al., 2015).”
The UU Submission to the Machine Translation Quality Estimation Task (2016, methodology): “In addition to the provided resources, further tools were used to extract the features: A modified version of the QuEst++ framework, (Specia et al., 2015) with processors and features added and modified where needed, used to extract the baseline features and a majority of our features.”
Recurrent Neural Network based Translation Quality Estimation (2016, methodology): “Shah et al. (2015) use continuous space language model features for sentence-level QE and word embedding features for word-level QE, in combination with other features produced by QuEst++ (Specia et al., 2015).”
Word embeddings and discourse information for Quality Estimation (2016, methodology): “Baseline We use the 17 Q U E ST ++ baseline features to train our baseline systems (Specia et al., 2015).”
The QT21/HimL Combined Machine Translation System (2016, methodology): “The system produces distinct 1000-best lists, for which we extend the feature set with the 17 baseline black-box features from sentence-level Quality Estimation (QE) produced with Quest++ 4 (Specia et al., 2015).”
USFD’s Phrase-level Quality Estimation Systems (2016, methodology): “The baseline phrase-level features provided by the organisers of the task are black-box features that were originally used for sentence-level quality estimation and extracted using the QuEst toolkit (Specia et al., 2015).”