A Unigram Orientation Model for Statistical Machine Translation

By Christoph Tillmann. North American Chapter of the Association for Computational Linguistics, 2004

337 citations recorded by the source corpus.

In this paper, we present a unigram segmentation model for statistical machine translation where the segmentation units are blocks: pairs of phrases without internal structure. The segmentation model uses a novel orientation component to handle swapping of neighbor blocks. During training, we collect block unigram counts with orientation: we count how often a block occurs to the left or to the right of some predecessor block. The orientation model is shown to improve translation performance over two models: 1) no block re-ordering is used, and 2) the block swapping is controlled only by a language model. We show experimental results on a standard Arabic-English translation task.

Subfields: Artificial Intelligence

View Paper: https://www.semanticscholar.org/paper/f2a1ea3f0308eb703433fc98d313c12bae7efdb2

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

Citations in the corpus

The RWTH Aachen German-English Machine Translation System for WMT 2015 (2015, methodology): “For the WMT 2015 German→English task, we applied discriminative maximum expected BLEU training as described by Wuebker et al. (2015). The expected BLEU objective function is optimized with the resilient backpropagation algorithm (RPROP) (Riedmiller and Braun, 1993).”
Combining Word Reordering Methods on different Linguistic Abstraction Levels for Statistical Machine Translation (2013, background): “Lexicalized re-ordering models (Koehn et al., 2005; Tillmann, 2004) include information about the orientation of adjacent phrases that is learned during phrase extraction.”
FBK@IWSLT 2011 (2011, methodology): “The decoder features a statistical log-linear model including one or more phrase translation models, target language models, a phrase reordering model [11, 12], distortion, word and phrase penalties.”
Phrasal: A Toolkit for New Directions in Statistical Machine Translation (2014, methodology): “The standard phrase orientation model (Tillmann, 2004) and the hierarchical model of Galley and Manning (2008) are available.”
Statistical Machine Translation (2014, neutral): “Statistical Machine Translation Dr Khalil Sima’an Motivation Challenges This course Word-Based Models Phrase-Based Models Limitations of PB Models Syntax Permutation factorization Synchronous Grammars Other issues (mostly covered later) Training phrase-based models Maximum-Likelihood problems Other ”
Continuous Space Translation Models with Neural Networks (2012, methodology): “In addition to the TM, fourteen feature functions are included: a target-language model; four lexicon models; six lexicalized reordering models (Tillmann, 2004; Crego et al., 2011); a distance-based distortion model; and finally a word-bonus model and a tuple-bonus model.”
LIMSI@WMT’16: Machine Translation of News (2016, methodology): “…13 conventional features are combined: 4 lexicon models similar to the ones used in standard phrase-based systems; 6 lexical-ized reordering models (Tillmann, 2004; Crego et al., 2011) aimed at predicting the orientation of the next translation unit; a “weak” distance-based distortion model ; and…”
A Simple and Effective Hierarchical Phrase Reordering Model (2008, methodology): “We compare our re-ordering model with related work (Tillman, 2004; Koehn et al., 2007) using a log-linear approach common to many state-of-theart statistical machine translation systems (Och and Ney, 2004).”
LIMSI @ WMT’14 Medical Translation Task (2014, methodology): “…13 conventional features are combined: 4 lexicon models similar to the ones used in standard phrase-based systems; 6 lexical-ized reordering models (Tillmann, 2004; Crego et al., 2011) aimed at predicting the orientation of the next translation unit; a “weak” distance-based distortion model ; and…”
The RWTH Aachen German-English Machine Translation System for WMT 2014 (2014, background): “The model scores monotone, swap, and discontinuous phrase orientations in the manner of the one presented by (Tillmann, 2004).”