A trainable transfer-based MT approach for languages with limited resources

What it is about

An approach for languages with few resources. Bilingual speakers, who need not be linguists, translate a designed English corpus into their language and mark word alignments in a special tool. The corpus is organised in minimal pairs and in order of complexity, from noun phrases to sentences, so that rules can be learned for each level. A learning module infers hierarchical syntactic transfer rules from the aligned data, and a run-time system applies them with a target decoder. The experiment was a Hindi-to-English system built in two months from a few thousand sentence pairs. The transfer system beat a statistical baseline under this small-data condition. A hand-written grammar beat the learned one, which beat no grammar, and the learned rules only covered noun phrases and prepositional phrases. The decoder's reordering window also changed the scores.

Ideas

I1. Minimal-pair probes for author deltas: for each author feature (thou and you, the LORD and Jehovah), build a small set of verse pairs that differ only in that feature, and use it to test whether a delta fixes exactly that. Count the tokens fixed and the ones changed by accident.
I2. Learn one level at a time: the paper learns noun phrases first and reports the gain per level. In the translator, learn the noun-phrase level of the encoder first and report its residual separately before any clause-level rules.
I3. Reordering window sweep: en/src/lin.rs orders words with Markov chains over classes, and a window over candidate orders could be swept like the paper's. Report the score per window, run twice, and check the result is identical.

For Mova

The minimal-pair elicitation is a cheap way to test deltas, and the level-by-level learning is a control for the claim that a rule is learned. The reordering sweep is a measurable knob in the linearizer. The paper's reliance on a few thousand elicited pairs is a design point Mova can match with the Bible pairs for the test of the author deltas. Its scores are from a different language pair and are not used.

Topics: Machine translation

Cite

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

@inproceedings{lavie2004trainable,
  title = {{A trainable transfer-based MT approach for languages with limited resources}},
  author = {Lavie, Alon and Probst, Katharina and Peterson, Erik and Vogel, Stephan and Levin, Lori and Font-Llitjos, Ariadna and Carbonell, Jaime},
  booktitle = {Proceedings of the 9th EAMT Workshop: Broadening horizons of machine translation and its applications},
  year = {2004},
  url = {https://aclanthology.org/2004.eamt-1.14/}
}

References

16 references, 2 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

Allen, J. 1995. Natural Language Understanding, Second Edition, Benjamin Cummings. find
Brown, P., Cocke, J., Della Pietra, V., Della Pietra, S., Jelinek, F., Lafferty, J., Mercer, R., and Roossin, P. 1990. A Statistical Approach to Machine Translation. Computational Linguistics 16(2), 79-85. find
Brown, P., Della Pietra, V., Della Pietra, S., and Mercer, R. 1993. The Mathematics of Statistical Machine Translation: Parameter Estimation. Computational Linguistics 19(2), 263-311. find
Brown, R. 1997. Automated Dictionary Extraction for Knowledge-free Example-based Translation. In Proceedings of International Conference on Theoretical and Methodological Issues in Machine Translation (TMI-1997); 111-118. find
Doddington, G. 2002. Automatic Evaluation of Machine Translation Quality using N-gram Cooccurrence Statistics. In Proceedings of Human Language Technologies Conference (HLT-2002); 128-132. find
Och, F. J. and Ney, H. Discriminative Training and Maximum Entropy Models for Statistical Machine Translation. In Proceedings of 40 th Annual Meeting of the Association for Computational Linguistics (ACL-2002). Philadelphia, PA. find
Papineni, K., Roukos, S., Ward, T. and Zhu, W. 2002. BLEU: a Method for Automatic Evaluation of Machine Translation. . In Proceedings of 40 th Annual Meeting of the Association for Computational Linguistics (ACL-2002). Philadelphia, PA. find our card
Papineni, K., Roukos, S., and Ward, T. 1998. Maximum Likelihood and Discriminative Training of Direct Translation Models. In Proceedings of the International Conference on Acoustics, Speech, and Signal Processing (ICASSP-98). 189-192. find
Peterson, E. 2002. Adapting a Transfer Engine for Rapid Machine Translation Development. M.S. Thesis, Georgetown University. find
Probst, K., Brown, R., Carbonell, J., Lavie, A., Levin, L., and Peterson, E. 2001. Design and Implementation of Controlled Elicitation for Machine Translation of Low-Density Languages. In Workshop MT2010 at Machine Translation Summit VIII. find
Probst, K. and Levin, L. 2002. Challenges in Automated Elicitation of a Controlled Bilingual Corpus. In Theoretical and Methodological Issues in Machine Translation 2002 (TMI-02). find
Probst, K., Levin, L., Peterson, E., Lavie, A., and Carbonell, J. 2003. MT for Resource-Poor Languages using Elicitation-based Learning of Syntactic Transfer Rules. Machine Translation. To appear. find
Sato, S. and Nagao, M. 1990. Towards Memory-based Translation. In Proceedings of COLING-90. 247-252. find
Vogel, S. and Tribble, A. 2002. Improving Statistical Machine Translation for a Speech-to-Speech Translation task. In Proceedings of the 7 th International Conference on Spoken Language Processing (ICSLP-02). find
Vogel, S., Zhang, Y., Tribble, A., Huang, F., Venugopal, A., Zhao, B., and Waibel, A. 2003. The CMU Statistical Translation System. In Proceedings of MT Summit IX, New Orleans, LA.; 402-409. find
Yamada, K. and Knight, K. 2001. A Syntax-based Statistical Translation Model. In Proceedings of the 39th Meeting of the Association for Computational Linguistics (ACL-01). find our card

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/lavie-2004-trainable-transfer-based-mt-approach/