Improving a Statistical MT System with Automatically Learned Rewrite Patterns

By Michael C. McCord. International Conference on Computational Linguistics, 2004

321 citations recorded by the source corpus.

Current clump-based statistical MT systems have two limitations with respect to word ordering: First, they lack a mechanism for expressing and using generalization that accounts for reorderings of linguistic phrases. Second, the ordering of target words in such systems does not respect linguistic phrase boundaries. To address these limitations, we propose to use automatically learned rewrite patterns to preprocess the source sentences so that they have a word order similar to that of the target language. Our system is a hybrid one. The basic model is statistical, but we use broad-coverage rule-based parsers in two ways - during training for learning rewrite patterns, and at runtime for reordering the source sentences. Our experiments show 10% relative improvement in Bleu measure.

Subfields: Artificial Intelligence

View Paper: https://www.semanticscholar.org/paper/4791cb189d2c07090051f39f2cb63bf4767faa3d

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

Yandex School of Data Analysis approach to English-Turkish translation at WMT16 News Translation Task (2016, methodology): “We directly apply English-to-Turkish reorderers described in Section 2.3 as a pre-processing step in the phrase-based MT pipeline, like e.g. (Xia and McCord, 2004; Collins et al., 2005).”
Syntactic Phrase Reordering for English-to-Arabic Statistical Machine Translation (2009, background): “(Xia and McCord, 2004) present a similar approach, with a notable difference: the re-ordering rules are automatically learned from aligning parse trees for both the source and target sentences.”
A Survey of Multilingual Neural Machine Translation (2019, background): “Pre-ordering can be done via hand-written rules [87, 120] or learnt rules/models [151, 162].”
Anaphora Models and Reordering for Phrase-Based SMT (2014, background): “Popović and Ney, 2006) or data-driven models (Xia and McCord, 2004; Genzel, 2010; Rottmann and Vogel, 2007; Niehues and Kolss, 2009) for preordering training data and system input have been explored in numerous publications.”
Natural Language Processing with Small Feed-Forward Networks (2017, methodology): “Preordering source-side words into the target-side word order is a useful preprocessing task for statistical machine translation (Xia and McCord, 2004; Collins et al., 2005; Nakagawa, 2015; de Gispert et al., 2015).”
Hierarchical Phrase-Based Translation (2007, unknown): “The same method has also been used by Probst et al. (2002) and Xia and McCord (2004) in conjunction with syntactic annotations to extract rules that are used for reordering prior to translation.”
Combining Word Reordering Methods on different Linguistic Abstraction Levels for Statistical Machine Translation (2013, methodology): “To our best knowledge Xia and McCord (2004) were the ﬁrst to model the word reordering problem as a preprocessing step.”
Experiments in Morphosyntactic Processing for Translating to and from German (2009, background): “…parse of the source sentence to change the word order to more closely match the word order of the target language have been studied by Niessen and Ney (2004), Xia and McCord (2004), Dr´abek and Yarowsky (2004), Collins et al. (2005), Popovi´c and Ney (2006), Wang et al. (2007) and many others.”
Investigating the Usefulness of Generalized Word Representations in SMT (2014, methodology): “Xia and McCord (2004) proposed a method to automatically learn rewrite rules to preorder source sentences.”
Clause Restructuring for Statistical Machine Translation (2005, methodology): “(Xia and McCord, 2004) describe an approach for translation from French to English, where reordering rules are acquired automatically.”