Exploiting Task Reversibility of DRS Parsing and Generation: Challenges and Insights from a Multi-lingual Perspective

What it is about

The paper studies error dynamics in two pipelines built on the same reversible task. Parse-Generate-Parse (PGP) takes text to DRS, back to text, and back to DRS; Generate-Parse-Generate (GPG) takes DRS to text, back to DRS, and back to text. Pre-trained byT5 models are used without retraining, on the Parallel Meaning Bank for English (high-resource), Italian (mid-resource) and Urdu (low-resource). Each example is labelled as improved (error corrected), unchanged or worsened (error amplified). The authors find that the pipelines more often propagate errors than correct them. Urdu is the most sensitive, Italian shows a moderate effect, and English is the most stable. Parsing scores are highest in English (SMATCH F1 93.56 without the pipeline, 93.06 with it).

Ideas

• I1. Use Mova's own stack for the same test: meta-world to text and back (GPG), and text to meta-world, to text and back (PGP). Label each example improved, unchanged or worse, as in the paper, and run twice for identical output.
• I2. Count error propagation by class: which divergence classes (lexical, construction, order, function word) grow across a round trip. A class that grows is the layer to fix first.
• I3. Language sensitivity: the paper finds the low-resource language most fragile. The Hebrew reader is the weakest part of the translator, so expect the same ordering there, and measure it with the same classes per source language.
• I4. Count-based augmentation: the paper multiplies its training data with name, lexical and grammatical substitutions. A cheap version on the Bible pairs swaps names and tenses by counts, then checks whether the round trip gets more stable. Measure on dev only.

For Mova

The round-trip test fits tools/roundtrip and tools/translate, with the classes the translator's residual already uses. The byT5 models are not for Mova. The finding that errors are more often propagated than corrected is a reason to measure the round trip before adding layers.

Topics: Machine translation

Cite

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

@inproceedings{amin2025exploiting,
  title = {{Exploiting Task Reversibility of DRS Parsing and Generation: Challenges and Insights from a Multi-lingual Perspective}},
  author = {Amin, Muhammad Saad and Anselma, Luca and Mazzei, Alessandro},
  booktitle = {Proceedings of the First Workshop on Language Models for Low-Resource Languages},
  year = {2025},
  url = {https://aclanthology.org/2025.loreslm-1.22/}
}

References

24 references, 6 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Lasha Abzianidze, Johannes Bjerva, Kilian Evang, Hessel Haagsma, Rik van Noord, Pierre Ludmann, Duc-Duy Nguyen, and Johan Bos. 2017. The Parallel Meaning Bank: Towards a multilingual corpus of translations annotated with compositional meaning representations. In Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 2, Short Papers, pages 242–247, Valencia, Spain. Association for Computational Linguistics. find our card
2. Johan Bos. 2021. Quantification annotation in discourse representation theory. In ISA 2021-17th Workshop on Interoperable Semantic Annotation, Groningen/Virtuel, Netherlands, June, pages 1–29. find
3. Johan Bos. 2023. The sequence notation: Catching complex meanings in simple graphs. In 15th International Conference on Computational Semantics, pages 195–208. Association for Computational Linguistics (ACL). find
4. Shu Cai and Kevin Knight. 2013. Smatch: an evaluation metric for semantic feature structures. In Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), pages 748–752. find
5. Federico Fancellu, Sorcha Gilroy, Adam Lopez, and Mirella Lapata. 2019. Semantic graph parsing with recurrent neural network DAG grammars. 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 2769–2778, Hong Kong, China. Association for Computational Linguistics. of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 397–415, Online. Association for Computational Linguistics. find
6. Rik van Noord. 2019. Neural boxer at the IWCS shared task on DRS parsing. in Proc. IWCS Shared Task on Semantic Parsing, Gothenburg, Sweden. Association for Computational Linguistics[. find
7. Kishore Papineni, Salim Roukos, Todd Ward, and Wei-Jing Zhu. 2002. Bleu: a method for automatic evaluation of machine translation. In Proceedings of the 40th annual meeting of the Association for Computational Linguistics, pages 311–318. find our card
8. Wessel Poelman, Rik van Noord, and Johan Bos. 2022. Transparent semantic parsing with universal dependencies using graph transformations. In 29th International Conference on Computational Linguistics, pages 4186–4192. Association for Computational Linguistics (ACL). find
9. Christiane Fellbaum. 1998. WordNet: An electronic lexical database. MIT press. find
10. Qiankun Fu, Yue Zhang, Jiangming Liu, and Meishan Zhang. 2020. DRTS parsing with structure-aware encoding and decoding. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 6818–6828, Online. Association for Computational Linguistics. find
11. Maja Popović. 2015. chrF: character n-gram F-score for automatic MT evaluation. In Proceedings of the Tenth Workshop on Statistical Machine Translation, pages 392–395, Lisbon, Portugal. Association for Computational Linguistics. find our card
12. Michael Hanna and Ondřej Bojar. 2021. A fine-grained analysis of BERTScore. In Proceedings of the Sixth Conference on Machine Translation, pages 507–517, Online. Association for Computational Linguistics. find
13. Kasia M Jaszczolt and Katarzyna Jaszczolt. 2023. Semantics, pragmatics, philosophy: a journey through meaning. Cambridge University Press. find
14. Ricardo Rei, Craig Stewart, Ana C Farinha, and Alon Lavie. 2020. COMET: A neural framework for MT evaluation. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 2685–2702, Online. Association for Computational Linguistics. find our card
15. Hans Kamp and Uwe Reyle. 1993. From Discourse to Logic: Introduction to Modeltheoretic Semantics of Natural Language, Formal Logic and Discourse Representation Theory. Kluwer Academic Publishers, Dordrecht. find
16. Lukas Stankevičius, Mantas Lukoševičius, Jurgita Kapočiūtė-Dzikienė, Monika Briedienė, and Tomas Krilavičius. 2022. Correcting diacritics and typos with a byt5 transformer model. Applied Sciences, 12(5):2636. find
17. Hans Kamp and Uwe Reyle. 2013. From discourse to logic: Introduction to modeltheoretic semantics of natural language, formal logic and discourse representation theory, volume 42. Springer Science & Business Media. find
18. Rik van Noord, Lasha Abzianidze, Hessel Haagsma, and Johan Bos. 2018. Evaluating scoped meaning representations. In Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018), Miyazaki, Japan. European Language Resources Association (ELRA). find
19. Hans Kamp, Josef Van Genabith, and Uwe Reyle. 2010. Discourse representation theory. In Handbook of Philosophical Logic: Volume 15, pages 125–394. Springer. find
20. Robert T Kasper. 1989. A flexible interface for linking applications to penman’s sentence generator. In Speech and Natural Language: Proceedings of a Workshop Held at Philadelphia, Pennsylvania, February 21-23, 1989. ↗
21. Chin-Yew Lin. 2004. ROUGE: A package for automatic evaluation of summaries. In Text Summarization Branches Out, pages 74–81, Barcelona, Spain. Association for Computational Linguistics. find
22. Jiangming Liu, Shay B. Cohen, and Mirella Lapata. 2021. Text generation from discourse representation structures. In Proceedings of the 2021 Conference Rik van Noord, Antonio Toral, and Johan Bos. 2019. Linguistic information in neural semantic parsing with multiple encoders. In Proc. 13th International Conference on Computational Semantics-Short Papers, pages 24–31. Association for Computational Linguistics (ACL). find our card
23. Chunliu Wang, Huiyuan Lai, Malvina Nissim, and Johan Bos. 2023. Pre-trained language-meaning models for multilingual parsing and generation. Preprint, arXiv:2306.00124. ↗
24. Chunliu Wang, Rik van Noord, Arianna Bisazza, and Johan Bos. 2021. Evaluating text generation from discourse representation structures. In Proceedings of the 1st Workshop on Natural Language Generation, Evaluation, and Metrics (GEM 2021), pages 73–83, Online. Association for Computational Linguistics. find our card

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/amin-2025-exploiting-task-reversibility-drs-parsing/