The jsRealB Text Realizer: Organization and Use Cases -- Revised version

What it is about

The paper describes the design of jsRealB, a text realiser for English and French that runs in a web page or as a Node module. Its input is a specification inspired by constituent syntax, with a dependency-style notation also available. The realiser takes care of agreement, conjugation, elision and word order, and it can build negative, passive and interrogative forms from one affirmative specification. The author shows that producing well-formed text is harder than it looks, because each language has its own details. The system has been used for data-to-text applications, for rebuilding literary sentences from Universal Dependency annotations, and in a Python port. The paper is a design report, and it gives many small examples of the rules at work.

Ideas

• I1. Variants as tree transformations. Derive the negative, passive and interrogative forms of an English tree by ordered transformations in en/src/lin.rs, and check on English UD sentences that each variant maps back to the original tree. A failed round trip is a counted error.
• I2. Use UD reproduction as an exact-mode test. The paper rebuilds sentences from UD annotations. For the English dev data, rebuild each sentence from its tree and count exact matches per class (word order, agreement, punctuation). This gives the residual of the surface step in bits.
• I3. Keep agreement and elision in separate, ordered rule tables in en/src/morph.rs, each built from counts and kept in BTreeMap order. For the translator's target side, the same split gives the function-word rules that open problem 4 needs.

For Mova

The paper fits en/src/lin.rs and en/src/morph.rs, the English realisation side. The variant-by-transformation idea and the round-trip test on UD sentences are the parts to borrow. French is outside the current scope, so only the design of the rule tables carries over.

Topics: Machine translation

Cite

The paper itself, not this card. BibTeX from the library's record of the work. arXiv 2012.15425

@misc{lapalme2020jsrealb,
  title = {{The jsRealB Text Realizer: Organization and Use Cases -- Revised version}},
  author = {Lapalme, Guy},
  year = {2020},
  eprint = {2012.15425},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2012.15425}
}

References

23 references, 5 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. John Bateman. KPML access page. http://purl.org/net/kpml, 2016. ↗
2. Guillaume Le Berre, Christophe Cerisara, Philippe Langlais, and Guy Lapalme. Unsupervised multiple choice question generation for out-of-domain Q&A fine-tuning. In ACL-2022, page 8p., Feb 2022. find
3. Daniel Braun, Kira Klimt, Daniela Schneider, and Florian Matthes. SimpleNLG-DE: Adapting SimpleNLG 4 to German. In Proceedings of the 12th International Conference on Natural Language Generation, pages 415–420, Tokyo, Japan, October–November 2019. Association for Computational Linguistics. find
4. Thiago Castro-Ferreira, Claire Gardent, Nikolai Ilinykh, Chris van der Lee, Simon Mille, Diego Moussalem, and Anastasia Shimorina. The 2020 Bilingual, Bi-Directional WebNLG+ Shared Task: Overview and Evaluation Results (WebNLG+ 2020). In Proceedings of the 3rd WebNLG Workshop on Natural Language Generation from the Semantic Web (WebNLG+ 2020), Dublin, Ireland (Virtual), 2020. Association for Computational Linguistics. find
5. Thiago Castro Ferreira, Chris van der Lee, Emiel van Miltenburg, and Emiel Krahmer. Neural data-to-text generation: A comparison between pipeline and end-to-end architectures. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), pages 552–562, Hong Kong, China, November 2019. Association for Computational Linguistics. find
7. Noam Chomsky. Syntactic structures. Mouton de Gruyter, 2nd edition, 2002. find
8. Nicolas Daoust and Guy Lapalme. JSreal: A Text Realizer for Web Programming. In Núria Gala, Reinhard Rapp, and Gemma Bel-Enguix, editors, Language Production, Cognition, and the Lexicon, number 6636 in Text, Speech and Language Technology, chapter 21, pages 363–378. Springer, jul 2014. find
9. Marie-Catherine de Marneffe, Christopher D. Manning, Joakim Nivre, and Daniel Zeman. Universal Dependencies. Computational Linguistics, 47(2):255–308, 07 2021. find our card
10. Michael Elhadad and Jacques Robin. An overview of SURGE: a reusable comprehensive syntactic realization component. In Eighth International Natural Language Generation Workshop (Posters and Demonstrations), 1996. find our card
11. Albert Gatt and Ehud Reiter. SimpleNLG: A realisation engine for practical applications. In Proceedings of the 12th European Workshop on Natural Language Generation (ENLG 2009), pages 90–93, Athens, Greece, March 2009. Association for Computational Linguistics. find
12. Guy Lapalme. Realizing Universal Dependencies structures using a symbolic approach. In Simon Mille, Anja Belz, Bernd Bohnet, Yvette Graham, and Leo Wanner, editors, The Second Multilingual Surface Realisation Shared Task (SR’19): Overview and Evaluation Results. In Proceedings of the 2nd Workshop on Multilingual Surface Realisation (MSR), (EMNLP-2019), page 8 pages, Hong-Kong, nov 2019. ACL. find
13. Guy Lapalme. Verbalizing AMR structures. http://rali.iro.umontreal.ca/rali/ sites/default/files/publis/GoPhi.pdf, Aug 2019. ↗
14. Guy Lapalme. Regenerating sentences from Universal Dependencies structures. http://rali.iro.umontreal.ca/JSrealB/current/demos/UDregenerator/ UDregenerator.pdf, Dec 2020. ↗
15. Guy Lapalme. Validation of Universal Dependencies by regeneration. In Proceedings of the Fifth Workshop on Universal Dependencies (UDW, SyntaxFest 2021), pages 109– 120, Sofia, Bulgaria, December 2021. Association for Computational Linguistics. find our card
16. François Lareau, Florie Lambrey, Ieva Dubinskaite, Daniel Galarreta-Piquette, and Maryam Nejat. GenDR: A generic deep realizer with complex lexicalization. In Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018), Miyazaki, Japan, May 2018. European Language Resources Association (ELRA). find our card
17. Benoit Lavoie and Owen Rainbow. A fast and portable realizer for text generation systems. In Fifth Conference on Applied Natural Language Processing, pages 265–268, Washington, DC, USA, March 1997. Association for Computational Linguistics. find our card
18. Susan W. Mcroy, Songsak Channarukul, and Syed S. Ali. An augmented template-based approach to text realization. Nat. Lang. Eng., 9(4):381–420, December 2003. find
19. Simon Mille, Roberto Carlini, Alicia Burga, and Leo Wanner. FORGe at SemEval-2017 task 9: Deep sentence generation based on a sequence of graph transducers. In Proceedings of the 11th International Workshop on Semantic Evaluation (SemEval-2017), pages 920–923, Vancouver, Canada, August 2017. Association for Computational Linguistics. find
20. Paul Molins and Guy Lapalme. JSrealB: A bilingual text realizer for web programming. In Proceedings of the 15th European Workshop on Natural Language Generation (ENLG), pages 109–111, Brighton, UK, September 2015. Association for Computational Linguistics. find
21. Lucien Tesnière. Éléments de syntaxe structurale. Librairie C. Klincksieck, 1959. find
22. Kees van Deemter, Emiel Krahmer, and Mariët Theune. Squibs and discussions: Real versus template-based natural language generation: A false opposition? Computational Linguistics, 31(1):15–24, 2005. find
23. Pierre-Luc Vaudry and Guy Lapalme. Adapting SimpleNLG for bilingual English-French realisation. In 14th European Conference on Natural Language Generation, pages 183–187, Sofia, Bulgaria, Aug 2013. find
24. Daniel Zeman, Joakim Nivre, Michell Abrams, and et al. Universal Dependencies 2.7. http://hdl.handle.net/11234/1-3424, nov 2020. ↗

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/lapalme-2020-jsrealb-text-realizer-organization-use/