Referential Translation Machines for Predicting Translation Quality

By Ergun Biçici, Andy Way. 2014

25 citations recorded by the source corpus.

We use referential translation machines (RTM) for quality estimation of translation outputs. RTMs are a computational model for identifying the translation acts between any two data sets with respect to interpretants selected in the same domain, which are effective when making monolingual and bilingual similarity judgments. RTMs achieve top performance in automatic, accurate, and language independent prediction of sentence-level and word-level statistical machine translation (SMT) quality. RTMs remove the need to access any SMT system specific information or prior knowledge of the training data or models used when generating the translations and achieve the top performance in WMT13 quality estimation task (QET13). We improve our RTM models with the Parallel FDA5 instance selection model, with additional features for predicting the translation performance, and with improved learning models. We develop RTM models for each WMT14 QET (QET14) subtask, obtain improvements over QET13 results, and rank 1st in all of the tasks and subtasks of QET14.

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

Using on-line available sources of bilingual information for word-level machine translation quality estimation (2015, methodology): “For example, the best system for the data set EN–ES (Camargo de Souza et al., 2014) used 163 features, while the winner system for the rest of data sets (Bic¸ici and Way, 2014; Bic¸ici, 2013) used 511,000 features.”
Multi-level Translation Quality Prediction with QuEst++ (2015, methodology): “We explore a range of features from recent work (Bicici and Way, 2014; Camargo de Souza et al., 2014; Luong et al., 2014; Wisniewski et al., 2014), totalling 40 features of seven types: Target context These are features that explore the context of the target word.”
Referential translation machines for predicting semantic similarity (2015, methodology): “…Workshop on Semantic Evaluation (Nakov and Zesch 2014; Nakov et al. 2015) in ﬁve semantic similarity tasks in total, and provide a comparison with the performance in bilingual settings such as the performance prediction of machine translation (Bic¸ici and Way 2014a; Bojar et al. 2014).”
Referential Translation Machines for Predicting Translation Quality and Related Statistics (2015, methodology): “…(SVR) with RBF (radial basis functions) kernel (Smola and Sch¨olkopf, 2004) for sentence and document translation prediction tasks and Global Linear Models (GLM) (Collins, 2002) with dynamic learning (GLMd) (Bic¸ici, 2013; Bic¸ici and Way, 2014) for word-level translation performance prediction.”
Target-Centric Features for Translation Quality Estimation (2014, methodology): “Finally, we would like to integrate referential translation machines (Bic¸ici, 2013; Bic¸ici and Way, 2014) into our system as they performed well in the WMT quality estimation tasks this and last year.”
Referential Translation Machines for Predicting Translation Performance (2016, methodology): “In Table 7, we list the RTM test results for tasks and subtasks that predict HTER or METEOR from QET16, QET15 (Bic¸ici et al., 2015b), QET14 (Bic¸ici and Way, 2014), and QET13 (Bic¸ici, 2013”