The DCU-ICTCAS MT system at WMT 2014 on German-English Translation Task

By Liangyou Li, Xiaofeng Wu, Santiago Cortes Vaíllo, Jun Xie, Andy Way, Qun Liu. 2014

9 citations recorded by the source corpus.

This paper describes the DCU submission to WMT 2014 on German-English translation task. Our system uses phrasebased translation model with several popular techniques, including Lexicalized Reordering Model, Operation Sequence Model and Language Model interpolation. Our final submission is the result of system combination on several systems which have different pre-processing and alignments.

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 2014 Workshop on Statistical Machine Translation (2014, methodology): “External resources are monolingual corpora taken from the WMT 2014 translation task for LMs, the MT system used for the inverse glass-box features (Li et al., 2014b) and, for error grammar parsing, the PennTreebank and an error grammar derived from it (Foster, 2007).”
On Using Very Large Target Vocabulary for Neural Machine Translation (2014, methodology): “As for English→German, the corpus was preprocessed, in a manner similar to (Peitz et al., 2014; Li et al., 2014), in order to remove many poorly translated sentences.”
Training RNNs as Fast as CNNs (2017, methodology): “Following standard practice (Peitz et al., 2014; Li et al., 2014; Jean et al., 2015), the training corpus was pre-processed and about 4 million translation pairs are left after processing.”
Simple Recurrent Units for Highly Parallelizable Recurrence (2017, methodology): “RNN takes 47 seconds per epoch. 4.4MACHINE TRANSLATION Dataset We use the WMT 2014 English!German translation task. We pre-process the training corpus following standard practice (Peitz et al., 2014; Li et al., 2014; Jean et al., 2015). About 4M translation pairs are left after processing. The news-”
Parallel FDA5 for Fast Deployment of Accurate Statistical Machine Translation Systems (2014, methodology): “FDA5 is also used for selecting the training set in the WMT14 medical translation task (Calixto et al., 2014) and the tuning set in the WMT14 German-English translation task (Li et al., 2014).”
An Empirical Study of Adequate Vision Span for Attention-Based Neural Machine Translation (2016, methodology): “The preprocessing procedure for De-En corpus is similar to Li et al. (2014), except we did not filter sentence pairs with language detection.”