Topological Ordering of Function Words in Hierarchical Phrase-based Translation

What it is about

Hierarchical phrase-based translation builds one kind of nonterminal, so it overgenerates reorderings. The authors propose a pairwise dominance model: for two neighbouring function words in a rule, it says which one sits higher in the derivation (leftFirst, rightFirst, dontCare or neither). The values are counted from maximal consistent alignments and used as a rule feature. Function words are not a linguistic list here but the N most frequent source words, N from 32 to 2048. On Chinese-English the model with N = 128 raised BLEU from 30.58 to 31.59 on MT06 and from 24.08 to 24.91 on MT08 (significant on MT06); on Arabic-English N = 128 gave 41.56 to 42.66 on MT06. Larger N adds content words and gains less, and N = 32 hurts.

Ideas

• I1. Dominance values for target function words. In lin.rs, for each pair of neighbouring function words, count from verse pairs which one sits higher in the target tree, keep the four values in a BTreeMap, and use them when the linearizer chooses between two orders of neighbouring particles. Measure the order-class residual on dev.
• I2. Frequency as the function-word set. Take the N most frequent tokens of each language's corpus as the function-word set, as the paper does, and compare it with the hand-written list. Report the residual for N = 32, 64, 128, 256, run twice.
• I3. A count threshold. Keep a dominance value only when its count passes a cut; a value seen rarely should fall back to dontCare, the paper's safe default.
• I4. An author diagnostic. Compare the dominance values across author pairs. Where two translators disagree on a pair, that pair is a candidate for an override in the author delta.

For Mova

The model is count-based and fits the translator's BTreeMap tables. It attacks the same order errors as problem 4 (function words as features) and problem 6 (archaic word order). The Chinese and Arabic results do not transfer to English-only work, so measure on Bible verses first.

Topics: Machine translation

Cite

The paper itself, not this card. BibTeX from the library's record of the work.

@inproceedings{setiawan2009topological,
  title = {{Topological Ordering of Function Words in Hierarchical Phrase-based Translation}},
  author = {Setiawan, Hendra and Kan, Min-Yen and Li, Haizhou and Resnik, Philip},
  booktitle = {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},
  year = {2009},
  url = {https://aclanthology.org/P09-1037/}
}

References

23 references, 4 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. David Chiang, Yuval Marton, and Philip Resnik. 2008. find
2. Online large-margin training of syntactic and structural translation features. In Proceedings of the 2008 Conference on Empirical Methods in Natural Language Processing, pages 224–233, Honolulu, Hawaii, October. find
3. David Chiang. 2005. A hierarchical phrase-based model for statistical machine translation. find our card
5. David Chiang. 2007. Hierarchical phrase-based translation. Computational Linguistics, 33(2):201–228. find our card
6. Jiaying Howard. 2002. A Student Handbook for Chinese Function Words. The Chinese University Press. find
7. Gonzalo Iglesias, Adria de Gispert, Eduardo R. Banga, and William Byrne. 2009. Rule ltering by pattern for efcient hierarchical translation. In Proceedings of the 12th Conference of the European Chapter of the Association of Computational Linguistics (to appear). find
8. R. Kneser and H. Ney. 1995. Improved backing-off for m-gram language modeling. find
10. Philipp Koehn, Franz J. Och, and Daniel Marcu. 2003. find
11. Statistical phrase-based translation. In Proceedings of the 2003 Human Language Technology Conference of the North American Chapter of the Association for Computational Linguistics, pages 127–133, Edmonton, Alberta, Canada, May. Association for Computational Linguistics. find
12. Philipp Koehn. 2004. Statistical signicance tests for In Proceedings of machine translation evaluation. find
13. EMNLP 2004, pages 388–395, Barcelona, Spain, July. find
14. Yuval Marton and Philip Resnik. 2008. Soft syntactic constraints for hierarchical phrased-based translation. find
15. In Proceedings of The 46th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies, pages 1003– 1011, Columbus, Ohio, June. Franz Josef Och and Hermann Ney. 2002. Discriminative training and maximum entropy models for statistical machine translation. find
17. Franz Josef Och and Hermann Ney. 2004. The alignment template approach to statistical machine translation. Computational Linguistics, 30(4):417–449. find
18. Kishore Papineni, Salim Roukos, Todd Ward, and Wei-Jing Zhu. a method for automatic 2002. Bleu: In Proceedings evaluation of machine translation. of 40th Annual Meeting of the Association for Computational Linguistics, pages 311–318, Philadelphia, Pennsylvania, USA, July. find
19. Hendra Setiawan, Min-Yen Kan, and Haizhou Li. 2007. Ordering phrases with function words. find
20. In Proceedings of the 45th Annual Meeting of the Association of Computational Linguistics, pages 712– 719, Prague, Czech Republic, June. Libin Shen, Jinxi Xu, and Ralph Weischedel. 2008. find
21. A new string-to-dependency machine translation algorithm with a target dependency language model. In Proceedings of The 46th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies, pages 577–585, Columbus, Ohio, June. David Vilar, Daniel Stein, and Hermann Ney. 2008. find our card
22. Analysing soft syntax features and heuristics for hierarchical phrase based machine translation. International Workshop on Spoken Language Translation 2008, pages 190–197, October. find
23. Dekai Wu. 1997. Stochastic inversion transduction grammars and bilingual parsing of parallel corpora. find our card
24. Computational Linguistics, 23(3):377–404, Sep. Tiejun Zhao, Yajuan Lv, Jianmin Yao, Hao Yu, Muyun Yang, and Fang Liu. 2001. Increasing accuracy of chinese segmentation with strategy of multi-step processing. find
25. Journal of Chinese Information Processing (Chinese Version), 1:13–18. Bowen Zhou, Bing Xiang, Xiaodan Zhu, and Yuqing Gao. 2008. Prior derivation models for formally syntax-based translation using linguistically syntactic parsing and tree kernels. find
27. Andreas Zollmann and Ashish Venugopal. 2006. Syntax augmented machine translation via chart parsing. find

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/setiawan-2009-topological-ordering-function-words-hierarchical/