Consistent Translation of Repeated Nouns using Syntactic and Semantic Cues

Abstract

We propose a method to decide whether two occurrences of the same noun in a source text should be translated consistently, i.e. using the same noun in the target text as well. We train and test classifiers that predict consistent translations based on lexical, syntactic, and semantic features. We first evaluate the accuracy of our classifiers intrinsically, in terms of the accuracy of consistency predictions, over a subset of the UN Corpus. Then, we also evaluate them in combination with phrase-based statistical MT systems for Chinese-to-English and Germanto-English. We compare the automatic post-editing of noun translations with the re-ranking of the translation hypotheses based on the classifiers' output, and also use these methods in combination. This improves over the baseline and closes up to 50% of the gap in BLEU scores between the baseline and an oracle classifier.

Document-Level Neural Machine Translation with Hierarchical Attention Networks

“…Statistical Machine Translation (SMT) Initial studies were based on cache memories (Tiede- mann, 2010; Gong et al, 2011). However, most of the work explicitly models discourse phenomena (Sim Smith, 2017) such as lexical cohesion (Meyer and Popescu-Belis, 2012;Xiong et al, 2013;Loáiciga and Grisot, 2016;Pu et al, 2017;Mascarell, 2017), coherence (Born et al, 2017), and coreference (Rios Gonzales and Tuggener, 2017;Miculicich Werlen and Popescu-Belis, 2017a). Hardmeier et al (2013) introduced the document-level SMT paradigm.…”

Lexical Chains meet Word Embeddings in Document-level Statistical Machine Translation

“…Tiedemann (2010) and Gong et al (2011) use cache-models for this purpose, and Xiao et al (2011) apply a three-steps procedure that consist of identifying the ambiguous words in a document, obtaining a set of consistent translation for each of them, and generating a new translation of the document, where the identified words are translated consistently. Pu et al (2017) also study consistency in translation and train classifiers on syntactic and semantic features to predict how to consistently translate pairs of nouns in a document. More specifically, Mascarell et al (2014) and Pu et al (2015) benefit from text dependencies to improve the translation of words that refer back to compounds.…”

Using Word Embeddings to Enforce Document-Level Lexical Consistency in Machine Translation

“…Another bilingual approach is presented in Martínez Garcia et al (2014b), where the resulting models are also evaluated in a cross-lingual lexical substitution task. Recently, WEs have been used in Pu et al (2017) to improve the consistency of noun translations by means of a postediting/re-ranking procedure with a phrase-based SMT system.…”