Developing an interlingual translation lexicon using WordNets and Grammatical Framework

What it is about

Grammatical Framework (GF) translates through an abstract syntax, and its usual lexicons give one target word per source word. This work builds a multi-sense, full-form interlingual lexicon from Princeton WordNet and the Universal WordNet mappings, with one abstract entry per synset and concrete lexicons for German, French, Finnish, Swedish, Hindi and Bulgarian. A word sense disambiguation tool then picks the sense in context. The evaluation is qualitative: a few example sentences per language, with and without disambiguation. The examples show better lexical choice in most cases. The remaining errors are a wrong preposition, a chemistry sense of the word base chosen for a numerical context, untranslated words in Hindi where lexicon coverage is low, and idioms rendered word by word. The authors name lexicon size and the noise of machine-generated mappings as the main limits, and they propose manual correction or parallel-corpus lexicons as next steps.

Ideas

• I1. Use the sense inventory as a veto on concept merges: two source words merge only when they share a synset. Merges such as be with have would be blocked by the check. Measure the merges with and without the veto by residual bits.
• I2. Choose among several target chunks of one concept by the context that matched most often, as the encoder already does, and test it against a disambiguation score on the verses where the rival chunks differ.
• I3. Replace the qualitative example list with a fixed set of verses and count the errors per class (function words, idioms, script, wrong sense). The paper's error list becomes a measure this way.

For Mova

The code is tools/translate/meta.rs and profile.rs. I1 addresses open problem 2 (merges that carry function words, such as unto_thou, and the be/have case). The paper's one-target-per-word default is the behaviour of the lexicon that the standard encoder replaces, so its failures are a useful list of cases to test on.

Topics: Word sense disambiguation · Machine translation

Cite

The paper itself, not this card. BibTeX from the library's record of the work. DOI 10.3115/v1/w14-5508

@inproceedings{virk2014developing,
  title = {{Developing an interlingual translation lexicon using WordNets and Grammatical Framework}},
  author = {Virk, Shafqat Mumtaz and Prasad, K.V.S. and Ranta, Aarne and Angelov, Krasimir},
  booktitle = {Proceedings of the Fifth Workshop on South and Southeast Asian Natural Language Processing},
  year = {2014},
  doi = {10.3115/v1/w14-5508},
  url = {https://doi.org/10.3115/v1/w14-5508}
}

References

29 references, 1 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Angelov, K. (2011). The Mechanics of the Grammatical Framework. PhD thesis, Chalmers University Of Technology. ISBN 978-91-7385-605-8. find
2. Angelov, K. and Enache, R. (2010). Typeful Ontologies with Direct Multilingual Verbalization. In Fuchs, N. and Rosner, M., editors, CNL 2010, Controlled Natural Language. find
3. Bringert, B., Hallgren, T., and Ranta., A. (2011). www.grammaticalframework.org/lib/doc/synopsis.html. find
4. Curry, H. B. (1961). Some logical aspects of grammatical structure. In Jakobson, R., editor, Structure of Language and its Mathematical Aspects: Proceedings of the Twelfth Symposium in Applied Mathematics, pages 56–68. American Mathematical Society. find
5. de Melo, G. and Weikum, G. (2009). Towards a Universal Wordnet by learning from combined evidence. In Proceedings of the 18th ACM Conference on Information and Knowledge Management (CIKM 2009), pages 513–522, New York, NY, USA. ACM. find
6. Delpech, E., Daille, B., Morin, E., and Lemaire, C. (2012). Extraction of domain-specific bilingual lexicon from comparable corpora: Compositional translation and ranking. In Proceedings of COLING 2012, pages 745–762, Mumbai, India. find
7. The COLING 2012 Organizing Committee. find
8. Détrez, G. and Ranta, A. (2012). Smart paradigms and the predictability and complexity of inflectional morphology. In EACL, pages 645–653. find
9. Dymetman, M., Lux, V., and Ranta, A. (2000). XML and multilingual document authoring: Convergent trends. In Proc. Computational Linguistics COLING, Saarbrücken, Germany, pages 243–249. International Committee on Computational Linguistics. find
10. Enache, R., España-Bonet, C., Ranta, A., and Márquez, L. (2012). A hybrid system for patent translation. In Proceedings of the 16th Annual Conference of the European Association for Machine Translation (EAMT12), Trento, Italy. find
11. Hindi-WordNet (2012). http://www.cfilt.iitb.ac.in. Hindi Wordnet. ↗
12. Lindén, K. and Carlson., L. (2010). Finnwordnet—wordnet på finska via översättning. coNordica—Nordic Journal of Lexicography, 17:119–140. find
13. Ljunglöf, P. (2004). The Expressivity and Complexity of Grammatical Framework. PhD thesis, Dept. of Computing Science, Chalmers University of Technology and Gothenburg University. http://www.cs.chalmers.se/~peb/pubs/p04-PhD-thesis.pdf. ↗
14. Lyons, J. (1968). Introduction to theoretical linguistics. Cambridge: Cambridge University Press. find
15. Martin-Löf, P. (1982). Constructive mathematics and computer programming. In Cohen, Los, Pfeiffer, and Podewski, editors, Logic, Methodology and Philosophy of Science VI, pages 153–175. North-Holland, Amsterdam. find
16. Miller, G. A. (1995). Wordnet: A lexical database for English. Communications of the ACM, 38:39–41. find our card
17. Montague, R. (1974). Formal Philosophy. Yale University Press, New Haven. Collected papers edited by Richmond Thomason. find
18. Prasad, K. V. S. and Shafqat, M. V. (2012). Computational evidence that Hindi and Urdu share a grammar but not the lexicon. In The 3rd Workshop on South and Southeast Asian NLP, COLING. find
19. Qian, L., Wang, H., Zhou, G., and Zhu, Q. (2012). Bilingual lexicon construction from comparable corpora via dependency mapping. In Proceedings of COLING 2012, pages 2275–2290, Mumbai, India. find
20. The COLING 2012 Organizing Committee. find
21. Ranta, A. (2004). Grammatical Framework: A Type-Theoretical Grammar Formalism. The Journal of Functional Programming, 14(2):145–189. http://www.cse.chalmers.se/~aarne/articles/gf-jfp. pdf. ↗
22. Ranta, A. (2011). Grammatical Framework: Programming with Multilingual Grammars. CSLI Publications, Stanford. ISBN-10: 1-57586-626-9 (Paper), 1-57586-627-7 (Cloth). GF resource grammar library synopsis. 2012. Universal Word—Hindi Lexicon. find
23. Lexi-Ranta, A. and Angelov, K. (2010). Implementing Controlled Languages in GF. In Proceedings of CNL-2009, Athens, volume 5972 of LNCS, pages 82–101. find
24. Ranta, A., Détrez, G., and Enache, R. (2012). Controlled language for everyday use: the MOLTO phrasebook. In CNL 2012: Controlled Natural Language, volume 7175 of LNCS/LNAI. find
25. Rosetta, M. T. (1994). Compositional Translation. Kluwer, Dordrecht. find
26. Seki, H., Matsumura, T., Fujii, M., and Kasami, T. (1991). On multiple context-free grammars. Theoretical Computer Science, 88:191–229. find
27. Shafqat, M., Humayoun, M., and Aarne, R. (2011). An open source Punjabi resource grammar. In Proceedings of the International Conference Recent Advances in Natural Language Processing 2011, pages 70–76, Hissar, Bulgaria. RANLP 2011 Organising Committee. http://aclweb.org/anthology/R11-1010. ↗
28. Stallman, R. (2001). Using and Porting the GNU Compiler Collection. Free Software Foundation. find
29. Zhong, Z. and Ng, H. T. (2010). It makes sense: A wide-coverage word sense disambiguation system for free text. In Proceedings of the ACL 2010 System Demonstrations, pages 78–83, Uppsala, Sweden. Association for Computational Linguistics. http://www.aclweb.org/anthology/P10-4014. ↗

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/virk-2014-developing-interlingual-translation-lexicon-wordnets/