A review of EBMT using proportional analogies

Some years ago a number of papers reported an experimental implementation of Example Based Machine Translation (EBMT) using Proportional Analogy. This approach, a type of analogical learning, was attractive because of its simplicity; and the papers reported considerable success with the method. This paper reviews what we believe to be the totality of research reported using this method, as an introduction to our own experiments in this framework, reported in a companion paper. We report first some lack of clarity in the previously published work, and then report our findings that the purity of the proportional analogy approach imposes huge run-time complexity for the EBMT task even when heuristics as hinted at in the original literature are applied to reduce the amount of computation.

[1]  Yves Lepage,et al.  The GREYC machine translation system for the IWSLT 2007 evaluation campaign , 2007, IWSLT.

[2]  Johanna D. Moore,et al.  36th Annual Meeting of the Association for Computational Linguistics and 17th International Conference on Computational Linguistics, COLING-ACL '98, August 10-14, 1998, Université de Montréal, Montréal, Quebec, Canada. Proceedings of the Conference. , 1998 .

[3]  François Yvon,et al.  Scaling-up analogical learning Apprentissage par analogie : passage à l'échelle , 2008 .

[4]  Chiori Hori,et al.  Overview of the IWSLT 2005 Evaluation Campaign , 2005, IWSLT.

[5]  Philippe Langlais,et al.  Translating Unknown Words by Analogical Learning , 2007, EMNLP.

[6]  Y. Lepage,et al.  The ‘purest’ EBMT System Ever Built: No Variables, No Templates, No Training, Examples, Just Examples, Only Examples , 2005, MTSUMMIT.

[7]  Yves Lepage,et al.  Lower and higher estimates of the number of “true analogies” between sentences contained in a large multilingual corpus , 2004, COLING.

[8]  Yves Lepage,et al.  ALEPH: an EBMT system based on the preservation of proportional analogies between sentences across languages , 2005, IWSLT.

[9]  Pierre Zweigenbaum,et al.  Improvements in Analogical Learning: Application to Translating Multi-Terms of the Medical Domain , 2009, EACL.

[10]  Yves Lepage,et al.  Purest ever example-based machine translation: Detailed presentation and assessment , 2005, Machine Translation.

[11]  Yves. Lepage,et al.  Analogies of form between chunks in Japanese are massive and far from being misleading , 2007 .

[12]  A. McMahon.  Historical Linguistics: Overview , 2001 .

[13]  François Yvon,et al.  Scaling up Analogical Learning , 2008, COLING.

[14]  Etienne Denoual.  Analogical translation of unknown words in a statistical machine translation framework , 2007, MTSUMMIT.

[15]  Yves Lepage,et al.  Solving Analogies on Words: An Algorithm , 1998, COLING-ACL.