Elamur Research⌕
☰
home / papers / 2026 / Align and Shine: Building High-quality Sentence-aligned Corpora for Multilingual Text Simplification

Align and Shine: Building High-quality Sentence-aligned Corpora for Multilingual Text Simplification

Summary

Supervised training and proper evaluation of text simplification systems require pairs of complex and simplified sentences. Outside English, which has ASSET and ParallelSEW, such open data are rare. The authors pair Vikidia articles, an encyclopedia written for children, with Wikipedia articles on the same topics in Catalan, English, French, Italian and Spanish. They then derive sentence-level links from these document pairs. The pipeline uses SentAlign, a hybrid aligner with three stages: it shortlists candidates by Gale and Church length ratios, fixes confident anchors through embedding similarity, and fills the remaining gaps with a Dijkstra shortest-path search that starts from the simple side. This design allows N-to-M links such as splits, merges and deletions. Three multilingual encoders (LaBSE, BGE-M3, SONAR) are compared with the Gale and Church and Hunalign baselines on a hand-aligned gold standard, and a similarity threshold is tuned for each language. The provided excerpt omits the numeric results, so the best encoder and its scores are not reported here. The released corpus is described as the first large open simplification resource for Catalan and Spanish. Because it comes from a single genre in every language, it also supports cross-lingual evaluation.

Key points

  • Source corpus: paired Wikipedia and Vikidia articles, matched by identical headings, with stub articles removed.
    • Document counts: Catalan 179, English 2,585, Spanish 3,875, Italian 3,902, French 33,438.
    • Vikidia has roughly 10–15 times fewer sentences than Wikipedia, except in French.
  • Gold standard: 15 randomly chosen document pairs per language, each aligned by hand without any pre-existing alignments by a single annotator for that language.
    • Accepted operations: lexical and syntactic simplification, splitting, summarization (N-to-1 and N-to-N), and deletion.
  • Filtering: pairs are kept only if their cosine similarity falls between a tuned lower bound τ and 0.95.
    • The 0.95 upper cap removes near-verbatim copies, a known weakness of WikiLarge-style alignments.
  • Evaluation: Precision, Recall and F1 are computed with the official SentAlign script in two modes.
    • Strict: only exact group matches count.
    • Lax: partial overlaps also earn credit.
  • Availability: the data are released at github.com/kenjihilasak/Align-and-Shine under CC-BY-SA. The work was funded by the Horizon Europe iDEM project.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

Fields: Machine translation · Natural language processing

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.63317/55pt8xqgkge6

@inproceedings{sanchez2026align,
  title = {{Align and Shine: Building High-quality Sentence-aligned Corpora for Multilingual Text Simplification}},
  author = {Sanchez, Luis Kenji Hilasaca and Khallaf, Nouran and Sharoff, Serge},
  booktitle = {Proceedings of the 19th Workshop on Building and Using Comparable Corpora (BUCC)},
  year = {2026},
  doi = {10.63317/55pt8xqgkge6},
  url = {https://doi.org/10.63317/55pt8xqgkge6}
}

References

21 references, 8 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Sisay Fissaha Adafre and Maarten de Rijke. 2006. Finding similar sentences across multiple languages in Wikipedia. In Proceedings of the Workshop on NEW TEXT Wikis and blogs and other dynamic text sources. find
  2. Fernando Alva-Manchego, Louis Martin, Antoine Bordes, Carolina Scarton, Benoît Sagot, and Lucia Specia. 2020. ASSET: A dataset for tuning and evaluation of sentence simplification models with multiple rewriting transformations. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 4668– 4679, Online. Association for Computational Linguistics. find our card
  3. Dominique Brunato, Andrea Cimino, Felice Dell’Orletta, and Giulia Venturi. 2016. PaCCSS-IT: A parallel corpus of complex-simple sentences for automatic text simplification. In Proc EMNLP, pages 351– 361, Austin, Texas. Association for Computational Linguistics. find
  4. Rémi Cardon and Natalia Grabar. 2020. French biomedical text simplification: When small and precise helps. In Proceedings of the 28th International Conference on Computational Linguistics, pages 710–716, Barcelona, Spain (Online). International Committee on Computational Linguistics. ↗
  5. William Coster and David Kauchak. 2011. Simple English Wikipedia: A new text simplification task. In Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies, pages 665–669, Portland, Oregon, USA. Association for Computational Linguistics. find our card
  6. Paul-Ambroise Duquenne, Holger Schwenk, and Benoît Sagot. 2023. Sonar: sentence-level multimodal and language-agnostic representations. arXiv preprint arXiv:2308.11466. ↗
  7. Fangxiaoyu Feng, Yinfei Yang, Daniel Cer, Naveen Arivazhagan, and Wei Wang. 2022. Language-agnostic BERT sentence embedding. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 878–891, Dublin, Ireland. Association for Computational Linguistics. find our card
  8. William A. Gale and Kenneth W. Church. 1993. A program for aligning sentences in bilingual corpora. Computational Linguistics, 19(1):75–102. ↗ our card
  9. Chao Jiang, Mounica Maddela, Wuwei Lan, Yang Zhong, and Wei Xu. 2020. Neural CRF model for sentence alignment in text simplification. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 7943–7960, Online. Association for Computational Linguistics. find our card
  10. Philipp Koehn. 2005. Europarl: A parallel corpus for statistical machine translation. In Proc. MT Summit X. find
  11. Christina Niklaus, Isabel Espinosa-Zaragoza, Víctor García-Muñoz, and Paloma Moreda. 2026. A comparative study of sentence alignment methods for Spanish text simplification. Language Resources and Evaluation, 60(2):29. ↗
  12. Michael J Ryan, Tarek Naous, and Wei Xu. 2023. Revisiting non-English text simplification: A unified multilingual benchmark. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 4898–4927, Toronto, Canada. Association for Computational Linguistics. find
  13. Horacio Saggion. 2017. Automatic text simplification. Synthesis Lectures on Human Language Technologies. Morgan & Claypool Publishers. ↗
  14. Holger Schwenk, Vishrav Chaudhary, Shuo Sun, Hongyu Gong, and Francisco Guzmán. 2019. WikiMatrix: Mining 135m parallel sentences in 1620 language pairs from Wikipedia. arXiv preprint arXiv:1907.05791. ↗ our card
  15. Serge Sharoff, Reinhard Rapp, and Pierre Zweigenbaum. 2023. Building and Using Comparable Corpora for Multilingual Natural Language Processing. Synthesis Lectures on Human Language Technologies. Springer Nature. ↗
  16. Sanja Štajner, Marc Franco-Salvador, Paolo Rosso, and Simone Paolo Ponzetto. 2018. CATS: A tool for customized alignment of text simplification corpora. In Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018). European Language Resources Association (ELRA). find
  17. Steinthor Steingrimsson, Hrafn Loftsson, and Andy Way. 2023. SentAlign: Accurate and scalable sentence alignment. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing: System Demonstrations, pages 256–263, Singapore. Association for Computational Linguistics. find
  18. D. Varga, P. Halacsy, A. Kornai, V. Nagy, L. Nemeth, and V. Tron. 2007. Parallel corpora for medium density languages. In Recent Advances in Natural Language Processing IV. Selected papers from RANLP-05, pages 247–258. John Benjamins. ↗
  19. Shitao Xiao, Zheng Liu, Peitian Zhang, Niklas Muennighoff, Defu Lian, and Jian-Yun Nie. 2024. C-pack: Packed resources for general chinese embeddings. In Proceedings of the 47th international ACM SIGIR conference on research and development in information retrieval, pages 641–649. find our card
  20. Tianyi Zhang, Varsha Kishore, Felix Wu, Kilian Weinberger, and Yoav Artzi. 2020. BERTScore: Evaluating text generation with BERT. In International Conference on Learning Representations. ↗
  21. Michał Ziemski, Marcin Junczys-Dowmunt, and Bruno Pouliquen. 2016. The United Nations parallel corpus v1.0. In Proc LREC, Portorož, Slovenia. find our card

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/sanchez-2026-align-shine-building-high-quality/