Search Descriptions

General

Neural machine Translation

Statistical Machine Translation

Search Publications


author

title

other

year

Refinements of Phrase Based Models

Phrase-based models have refined in many ways. We discuss in detail reordering models, pruning and smoothing methods, and the inclusion of context features in separate sections elsewhere, and address here other refinements.

Phrase Based Model Refinement is the main subject of 27 publications. 4 are discussed here.

Topics in PhraseBasedModels

Phrase Based Model Training | Phrase Based Model EM | Phrase Based Model Refinement | Context Features | Smoothing | Paraphrasing | Pruning Large Models | Suffix Arrays | Reordering Models | Phrase Based Vs EBMT

Publications

Usually, the segmentation of the source is not modeled, or only a phrase count feature is used, but adding a source phrase segmentation model may be beneficial (Blackwood et al., 2008). When using automatically learned segmentation boundaries these may outperform linguistically motivated constituent boundaries (Xiong et al., 2010).
Models may allow word insertion to account for spurious function words (Xu, 2005), or allow for words to be dropped by translating them into the empty phrase (Li et al., 2008).

Benchmarks

Discussion

Related Topics

New Publications

  • Durrani et al. (2015)
  • Thu et al. (2016)
  • Huck et al. (2013)
  • Feng et al. (2014)
  • Durrani et al. (2013)
  • Durrani et al. (2013)
  • Servan and Petitrenaud (2012)
  • Feng et al. (2010)
  • Huck and Ney (2012)
  • Le et al. (2012)
  • Costa-jussà and Fonollosa (2005)
  • Durrani et al. (2011)
  • Mansour and Ney (2012)
  • Finch and Sumita (2009)
  • Yahyaei and Monz (2009)
  • Liberato et al. (2010)
  • Ling et al. (2010)
  • Saers and Wu (2011)
  • Zbib et al. (2010)
  • Cao et al. (2010)
  • Lavergne et al. (2011)
  • Costa-jussà et al. (2007)
  • Matveeva and Levow (2006)

Actions

Download

Contribute