A decoding method of system combination based on hypergraph in SMT

Authors:
Yupeng Liu;Sheng Li;Tiejun Zhao
Affiliations:
Machine Transaltion Lab, Harbin Institute of Technology, Nangang, Harbin, China;Machine Transaltion Lab, Harbin Institute of Technology, Nangang, Harbin, China;Machine Transaltion Lab, Harbin Institute of Technology, Nangang, Harbin, China
Venue:
AICI'11 Proceedings of the Third international conference on Artificial intelligence and computational intelligence - Volume Part III
Year:
2011

Citing 17
Cited 0

Is it harder to parse Chinese, or the Chinese Treebank?

ACL '03 Proceedings of the 41st Annual Meeting on Association for Computational Linguistics - Volume 1
Maximum entropy based phrase reordering model for statistical machine translation

ACL-44 Proceedings of the 21st International Conference on Computational Linguistics and the 44th annual meeting of the Association for Computational Linguistics
Hierarchical Phrase-Based Translation

Computational Linguistics
Machine translation system combination using ITG-based alignments

HLT-Short '08 Proceedings of the 46th Annual Meeting of the Association for Computational Linguistics on Human Language Technologies: Short Papers
Indirect-HMM-based hypothesis alignment for combining outputs from machine translation systems

EMNLP '08 Proceedings of the Conference on Empirical Methods in Natural Language Processing
Using dependency order templates to improve generality in translation

StatMT '07 Proceedings of the Second Workshop on Statistical Machine Translation
Mixture-model adaptation for SMT

StatMT '07 Proceedings of the Second Workshop on Statistical Machine Translation
Better k-best parsing

Parsing '05 Proceedings of the Ninth International Workshop on Parsing Technology
Efficient Minimum Error Rate Training and Minimum Bayes-Risk decoding for translation hypergraphs and lattices

ACL '09 Proceedings of the Joint Conference of the 47th Annual Meeting of the ACL and the 4th International Joint Conference on Natural Language Processing of the AFNLP: Volume 1 - Volume 1
Fast consensus decoding over translation forests

ACL '09 Proceedings of the Joint Conference of the 47th Annual Meeting of the ACL and the 4th International Joint Conference on Natural Language Processing of the AFNLP: Volume 2 - Volume 2
Joint decoding with multiple translation models

ACL '09 Proceedings of the Joint Conference of the 47th Annual Meeting of the ACL and the 4th International Joint Conference on Natural Language Processing of the AFNLP: Volume 2 - Volume 2
Variational decoding for statistical machine translation

ACL '09 Proceedings of the Joint Conference of the 47th Annual Meeting of the ACL and the 4th International Joint Conference on Natural Language Processing of the AFNLP: Volume 2 - Volume 2
Incremental HMM alignment for MT system combination

ACL '09 Proceedings of the Joint Conference of the 47th Annual Meeting of the ACL and the 4th International Joint Conference on Natural Language Processing of the AFNLP: Volume 2 - Volume 2
First- and second-order expectation semirings with applications to minimum-risk training on translation forests

EMNLP '09 Proceedings of the 2009 Conference on Empirical Methods in Natural Language Processing: Volume 1 - Volume 1
Lattice-based system combination for statistical machine translation

EMNLP '09 Proceedings of the 2009 Conference on Empirical Methods in Natural Language Processing: Volume 3 - Volume 3
Joint optimization for machine translation system combination

EMNLP '09 Proceedings of the 2009 Conference on Empirical Methods in Natural Language Processing: Volume 3 - Volume 3
Model combination for machine translation

HLT '10 Human Language Technologies: The 2010 Annual Conference of the North American Chapter of the Association for Computational Linguistics

Quantified Score

Hi-index	0.00

Visualization

Abstract

The word level system combination, which is better than phrase level and sentence level, has emerged as a powerful post-processing method for statistical machine translation (SMT). This paper first give the definition of HyperGraph(HG) as a kind of compact data structure in SMT, and then introduce simple bracket transduction grammar(SBTG) for hypergraph decoding. To optimize the more feature weights, we introduce minimum risk (MR) with deterministic annealing (DA) into the training criterion, and compare two classic training procedures in experiment. The deoding approaches of ngram model based on hypergraph are shown to be superior to conventional cube pruning in the setting of the Chinese-to-English track of the 2008 NIST Open MT evaluation.