The Apertium machine translation platform: Five years on

What it is about

A retrospective of Apertium after five years: how a consortium of Spanish universities and companies turned two inherited MT systems into a free engine with data for many language pairs. The paper sets out the design philosophy: start from word-for-word translation and add lexical processing, tagging and local reordering rules; keep the engine separate from the language data; use copyleft so that the shared data stays shared. It describes the modular pipeline, a developer community of more than a hundred people, the research built on the platform, and the business models around it. It ends with the open challenges, chiefly extending a shallow-transfer design from closely related languages to distant ones. No new measurement is reported.

Ideas

• I1. Make word-for-word the explicit floor of the translator. In tools/translate every layer should be reported against a per-author word-for-word baseline, so a gain from the meta-world or the encoder is a gain over something that is measured the same way.
• I2. Treat reproducibility as a gate, not a remark. Each published translator number should be rebuildable from the data files and the seed alone, in a fresh checkout, with two identical runs. The paper's argument for free software is the same argument behind Mova's determinism rule.
• I3. Track the cost of distance. Apertium needed extra machinery for less related pairs. For the Hebrew reader, measure how the error splits by distance between the source tree and the meta-world (for example by the number of relations that change head), not only by a single score.

For Mova

The code is tools/translate (hebrew.rs, meta.rs, measure.rs) and tools/roundtrip. The word-for-word floor belongs in measure.rs; the distance split belongs with the Hebrew reader, which is the weakest part of the translator. The shallow-transfer architecture is not for Mova, because Mova works from trees and frames, but the reproducibility and baseline habits carry over directly.

Topics: Machine translation

Cite

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

@inproceedings{forcada2009apertium,
  title = {{The Apertium machine translation platform: Five years on}},
  author = {Forcada, Mikel L. and Tyers, Francis M. and Ramírez-Sánchez, Gema},
  booktitle = {Proceedings of the First International Workshop on Free/Open-Source Rule-Based Machine Translation},
  year = {2009},
  url = {https://aclanthology.org/2009.freeopmt-1.3/}
}

References

26 references, 3 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Alegria, I., Diaz de Ilarraza, A., Labaka, G., Lersundi, M., Mayor, A., Sarasola, K., Forcada, M. L., Ortiz-Rojas, S., and Padró, L. (2005). An open architecture for transfer-based mt between spanish and basque. In OSMaTran, A workshop at MT Summit X (Phuket, Thailand, September 12–16, 2005), pages 12–16. 23 http://apertium.svn.sf.net/viewvc/apertium/ ↗
2. Armentano-Oller, C., Carrasco, R. C., Corbı́-Bellot, A. M., Forcada, M. L., Ginestı́-Rosell, M., Ortiz-Rojas, S., Pérez-Ortiz, J. A., Ramı́rez-Sánchez, G., Sánchez-Martı́nez, F., and Scalco, M. A. (2006). Open-source Portuguese–Spanish machine translation. In Vieira, R., Quaresma, P., Nunes, M., Mamede, N., Oliveira, C., and Dias, M., editors, Computational Processing of the Portuguese Language, Proc. PROPOR 2006, volume 3960 of LNCS, pages 50–59. Springer-Verlag. find
3. Armentano-Oller, C., Corbı́-Bellot, A. M., Forcada, M. L., Ginestı́-Rosell, M., Mon-tava Belda, M. A., Ortiz-Rojas, S., Pérez-Ortiz, J. A., Ramı́rez-Sánchez, G., and Sánchez-Martı́nez, F. (2007). Apertium, una plataforma de código abierto para el desarrollo de sistemas de traducción automática. In Rodrı́guez Galván, J. R. and Palomo Duarte, M., editors, Proc. of the FLOSS International Conf. 2007, pages 5–20. Servicio de Publicaciones de la Universidad de Cadiz. find
4. Armentano-Oller, C. and Forcada, M. (2008). Re-utilización de datos linguısticos para la crea-cion de un sistema de traduccion automatica para un nuevo par de lenguas. PLN, 41:243– 250. find
5. Canals-Marote, R., Esteve-Guillen, A., Garrido-Alenda, A., Guardiola-Savall, M., Iturraspe-Bellver, A., Montserrat-Buendia, S., Ortiz-Rojas, S., Pastor-Pina, H., Perez-Antón, P., and Forcada, M. (2001). The Spanish–Catalan machine translation system interNOSTRUM. In Proc. of MT Summit VIII. Santiago de Compostela, Spain, 18–22 July 2001. find
6. Carme Armentano-Oller, C., Corbı́-Bellot, A. M., Forcada, M. L., Ginestı́-Rosell, M., Bonev, B., Ortiz-Rojas, S., Pérez-Ortiz, J. A., Ramı́rez-Sánchez, G., and Sánchez-Martı́nez, F. (2005). An open-source shallow-transfer machine translation toolbox: consequences of its release and availability. In OSMaTran, A workshop at MT Summit X, pages 23–30. find
7. Carreras, X., Chao, I., Padro, L., and Padro, M. (2004). Freeling: An open-source suite of language analyzers. In Proc. of the 4th LREC, volume 4. find
8. Caseli, H., Nunes, M., and Forcada, M. (2006). Automatic induction of bilingual resources from aligned parallel corpora: application to shallow-transfer machine translation. MT, 20(4):227–245. ↗
9. Caseli, H. M. and Nunes, M. G. V. (2006). Automatic transfer rule induction from parallel corpora. In Proc. of the 3rd Workshop on MSc dissertations and PhD theses in AI (WTDIA) - International Joint Conf. IBERAMIA/SBIA/SBRN 2006, pages 1–10. find
10. Corbı́-Bellot, A. M., Forcada, M. L., Ortiz-Rojas, S., Pérez-Ortiz, J. A., Ramı́rez-Sánchez, G., Sánchez-Martı́nez, F., Alegria, I., Mayor, A., and Sarasola, K. (2005). An open-source shallow-transfer machine translation engine for the romance languages of spain. In Proc. of the Tenth Conf. of the European Association for MT, pages 79–86. find
11. Garrido-Alenda, A., Gilabert Zarco, P., Pérez-Ortiz, J. A., Pertusa-Ibáñez, A., Ramı́rez-Sánchez, G., Sánchez-Martı́nez, F., Scalco, M. A., and Forcada, M. L. (2004). Shallow parsing for Portuguese–Spanish machine translation. In Branco, A., Mendes, A., and Ribeiro, R., editors, Language technology for Portuguese: shallow processing tools and resources, pages 135–144. Edições Colibri, Lisboa. find
12. Ginestı́-Rosell, M., Ramı́rez-Sánchez, G., Ortiz-Rojas, S., Tyers, F. M., and Forcada, M. L. (2009). Development of a free Basque to Spanish machine translation system. PLN, (43):187–195. find
13. Karlsson, F. (1995). Constraint Grammar: a language-independent system for parsing unrestricted text. Walter de Gruyter. find
14. McIlroy, M., Pinson, E., and Tague, B. (1978). Unix time-sharing system forward. The Bell System Technical J., 57(6 part 2):1902. find
15. Pedersen, T. (2008). Empiricism is not a matter of faith. Comp. Ling., 34(3):465–470. find
16. Ramı́rez-Sánchez, G., Sánchez-Martı́nez, F., Ortiz-Rojas, S., Pérez-Ortiz, J. A., and Forcada, M. L. (2006). Opentrad Apertium open-source machine translation system: an opportunity for business and research. In Proc. of Translating and the Computer 28 Conf. find our card
17. Raymond, E. S. (2004). The Art of Unix Programming. Addison-Wesley. find
18. Sánchez-Martı́nez, F. (2008). Using unsupervised corpus-based methods to build rule-based machine translation systems. PhD thesis, Universitat d’Alacant. find
19. Sánchez-Martı́nez, F. and Forcada, M. L. (2007). Automatic induction of shallow-transfer rules for open-source machine translation. In Way, A. and Gawronska, B., editors, Proc. of the 11th Conf. of TMI (TMI 2007), volume 1, pages 181–190. Skövde University Studies in Informatics. find our card
20. Sánchez-Martı́nez, F. and Forcada, M. L. (2009). Inferring shallow-transfer machine translation rules from small parallel corpora. J. of AI Research, 34:605–635. find our card
21. Sánchez-Martı́nez, F., Pérez-Ortiz, J. A., and Forcada, M. L. (2005). Target-language-driven agglomerative part-of-speech tag clustering for machine translation. In Proc. of the International Conf. RANLP - 2005 (Recent Advances in NLP), pages 471–477. find
22. Sánchez-Martı́nez, F., Pérez-Ortiz, J. A., and Forcada, M. L. (2006). Speeding up target-language driven part-of-speech tagger training for machine translation. LNCS, 4293:844–854. find
23. Sánchez-Martı́nez, F., Pérez-Ortiz, J. A., and Forcada, M. L. (2008). Using target-language information to train part-of-speech taggers for machine translation. MT, 22(1-2):29–66. find
24. Tyers, F. and Donnelly, K. (2009). apertium-cy– a collaboratively-developed free RBMT system for Welsh to English. Prague Bull. of Math. Ling., 91:57–66. ↗
25. Tyers, F. M. (2009). Rule-based augmentation of training data in breton–french statistical machine translation. In Proc. of the 13th Annual Conf. of the European Association of MT, EAMT09, pages 213–218. find
26. Tyers, F. M., Wiechetek, L., and Trosterud, T. (2009). Developing prototypes for machine translation between two Sámi languages. In Proc. of the 13th Annual Conf. of the EAMT, EAMT09, pages 120–128. find

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/forcada-2009-apertium-machine-translation-platform-five/