Insertion and deletion models for statistical machine translation

  • Authors:

  • Matthias Huck;Hermann Ney

  • Affiliations:

  • RWTH Aachen University, Aachen, Germany;RWTH Aachen University, Aachen, Germany

  • Venue:
  • NAACL HLT '12 Proceedings of the 2012 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

We investigate insertion and deletion models for hierarchical phrase-based statistical machine translation. Insertion and deletion models are designed as a means to avoid the omission of content words in the hypotheses. In our case, they are implemented as phrase-level feature functions which count the number of inserted or deleted words. An English word is considered inserted or deleted based on lexical probabilities with the words on the foreign language side of the phrase. Related techniques have been employed before by Och et al. (2003) in an n-best reranking framework and by Mauser et al. (2006) and Zens (2008) in a standard phrase-based translation system. We propose novel thresholding methods in this work and study insertion and deletion features which are based on two different types of lexicon models. We give an extensive experimental evaluation of all these variants on the NIST Chinese→English translation task.