Elamur Research⌕
☰
home / papers / 2026 / AC/DG at Cruciverb-IT: Retrieval-Based Approaches for Italian Crossword Clue Answering

AC/DG at Cruciverb-IT: Retrieval-Based Approaches for Italian Crossword Clue Answering

Paper · 2026 · Ali Yassine, Claudio Savelli, Davide Napolitano et al.

Summary

The paper tackles Task 1 of Cruciverb-IT (EVALITA 2026): given an Italian crossword clue and the length of the expected answer, a system must return a ranked list of candidate solutions. Italian makes this hard because it has many near-synonyms and rich inflection, such as gender and number variants. The AC/DG team (Politecnico di Torino) frames the problem as retrieval over a training set of roughly 410,000 clue–answer pairs built from ItaCW plus web-collected data. They compare three systems:

  • a tuned BM25 lexical retriever;
  • a dense retriever built on Italian sentence embeddings;
  • a hybrid in which a zero-shot Qwen3 8B model judges and reorders the top BM25 and dense candidates, and can generate a new answer if neither seems right.

All three beat the official BM25 baseline. Dense retrieval performs best on every metric. The LLM judge does not surpass it, partly because near-duplicate clues produce plausible but wrong distractors. The work matters because it shows that simple embedding-based retrieval is a strong, inexpensive approach for crossword answering in a morphologically rich, non-English language.

Key points

  • Data and evaluation: The dataset combines ItaCW with web-sourced clue–answer pairs (about 410k unique pairs). External sources containing crossword clues were banned. Systems are scored with Acc@1, Acc@10 and MRR.
  • Official baseline: It retrieves the 10 most BM25-similar training clues and returns their answers. Scores were Acc@1 0.394, Acc@10 0.612 and MRR 0.457.
  • Dense retrieval (best system): It uses nickprock/sentence-bert-base-italian-xxl-uncased and reaches Acc@1 0.510 (+29.4%), Acc@10 0.730 and MRR 0.570.
  • Optimized BM25: It scores Acc@1 0.470, Acc@10 0.670 and MRR 0.530. It is strongest when test clues lexically overlap with training clues.
  • Hybrid Qwen3 8B judge (zero-shot, with generative fallback): It scores Acc@1 0.460, Acc@10 0.690 and MRR 0.520. That puts it below BM25 at Acc@1 but above it at Acc@10. The authors propose future work on per-category clue analysis and on full grid solving (Task 2).

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

 Institutions

🎓Politecnico di TorinoTurin · Italy11 works · 18 citations

 Fields, topics and concepts

📚 Shelves 5The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 worksInformation Systems · 1,437 works

🧭 Topics 1Research subjects and languages the work is about.Information retrieval & search · 473 works

Cite

Cite as Ali Yassine, Claudio Savelli, Davide Napolitano, Giuseppe Gallipoli, Luca Cagliero and Elena Baralis. 2026. AC/DG at Cruciverb-IT: Retrieval-Based Approaches for Italian Crossword Clue Answering. Proceedings of the Ninth Evaluation Campaign of Natural Language Processing and Speech Tools for Italian. Final Workshop (EVALITA 2026).

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

@inproceedings{yassine2026ac,
  title = {{AC/DG at Cruciverb-IT: Retrieval-Based Approaches for Italian Crossword Clue Answering}},
  author = {Ali Yassine, Claudio Savelli, Davide Napolitano, Giuseppe Gallipoli, Luca Cagliero and Elena Baralis},
  booktitle = {Proceedings of the Ninth Evaluation Campaign of Natural Language Processing and Speech Tools for Italian. Final Workshop (EVALITA 2026)},
  year = {2026},
  url = {https://aclanthology.org/2026.evalita-1.37/}
}

References

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

  1. S. Kulshreshtha, O. Kovaleva, N. Shivagunde, A. Rumshisky, Down and across: Introducing crossword-solving as a new NLP benchmark, in: S. Muresan, P. Nakov, A. Villavicencio (Eds.), Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), Associ… find
  2. G. Angelini, M. Ernandes, M. Gori, Solving italian crosswords using the web, in: S. Bandini, S. Manzoni (Eds.), AI*IA 2005: Advances in Artificial Intelligence, Springer Berlin Heidelberg, Berlin, Heidelberg, 2005, pp. 393–405. find
  3. G. Barlacchi, M. Nicosia, A. Moschitti, A retrieval model for automatic resolution of crossword puzzles in italian language, in: Proceedings of the First Italian Conference on Computational Linguistics CLiC-it 2014 & the Fourth International Workshop EVALITA 2014: 9-11 December 2014, Pisa, Pisa Univ… find
  4. P. Giadikiaroglou, M. Lymperaiou, G. Filandrianos, G. Stamou, Puzzle solving using reasoning of large language models: A survey, in: Y. Al-Onaizan, M. Bansal, Y.-N. Chen (Eds.), Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, Association for Computational Ling…
  5. S. Saha, S. Chakraborty, S. Saha, U. Garain, Language models are crossword solvers, in: L. Chiruzzo, A. Ritter, L. Wang (Eds.), Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Pape… find
  6. A. Sadallah, D. Kotova, E. Kochmar, What makes cryptic crosswords challenging for LLMs?, in: O. Rambow, L. Wanner, M. Apidianaki, H. Al-Khalifa, B. D. Eugenio, S. Schockaert (Eds.), Proceedings of the 31st International Conference on Computational Linguistics, Association for Computational Linguisti… find
  7. C. Ciaccio, G. Sarti, A. Miaschi, F. Dell’Orletta, Crossword space: Latent manifold learning for Italian crosswords and beyond, in: C. Bosco, E. Jezek, M. Polignano, M. Sanguinetti (Eds.), Proceedings of the Eleventh Italian Conference on Computational Linguistics (CLiC-it 2025), CEUR Workshop Proce… find
  8. C. Ciaccio, G. Sarti, A. Miaschi, F. Dell’Orletta, M. Nissim, Cruciverb-it @ evalita 2026: Overview of the crossword solving in italian task, in: Proceedings of the Ninth Evaluation Campaign of Natural Language Processing and Speech Tools for Italian. Final Workshop (EVALITA 2026), CEUR.org, Bari, I… find
  9. F. Cutugno, A. Miaschi, A. P. Aprosio, G. Rambelli, L. Siciliani, M. A. Stranisci, Evalita 2026: Overview of the 9th evaluation campaign of natural language processing and speech tools for italian, in: Proceedings of the Ninth Evaluation Campaign of Natural Language Processing and Speech Tools for I… find
  10. E. Wallace, N. Tomlin, A. Xu, K. Yang, E. Pathak, M. Ginsberg, D. Klein, Automated crossword solving, in: S. Muresan, P. Nakov, A. Villavicencio (Eds.), Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), Association for Computational Ling… find
  11. J. Rozner, C. Potts, K. Mahowald, Decrypting cryptic crosswords: Semantically complex wordplay puzzles as a target for nlp, in: M. Ranzato, A. Beygelzimer, Y. Dauphin, P. Liang, J. W. Vaughan (Eds.), Advances in Neural Information Processing Systems, volume 34, Curran Associates, Inc., 2021, pp. 114… find
  12. A. Efrat, U. Shaham, D. Kilman, O. Levy, Cryptonite: A cryptic crossword benchmark for extreme ambiguity in language, in: M.-F. Moens, X. Huang, L. Specia, S. W.-t. Yih (Eds.), Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, Association for Computational Lingu… find
  13. J. Huang, K. C.-C. Chang, Towards reasoning in large language models: A survey, in: A. Rogers, J. Boyd-Graber, N. Okazaki (Eds.), Findings of the Association for Computational Linguistics: ACL 2023, Association for Computational Linguistics, Toronto, Canada, 2023, pp. 1049–1065. URL: https://aclanth…
  14. G. Sarti, T. Caselli, M. Nissim, A. Bisazza, Non verbis, sed rebus: Large language models are weak solvers of Italian rebuses, in: F. Dell’Orletta, A. Lenci, S. Montemagni, R. Sprugnoli (Eds.), Proceedings of the Tenth Italian Conference on Computational Linguistics (CLiC-it 2024), CEUR Workshop Pro… find
  15. K. Zeinalipour, T. Iaquinta, A. Zanollo, G. Angelini, L. Rigutini, M. Maggini, M. Gori, Italian crossword generator: Enhancing education through interactive word puzzles, in: F. Boschetti, G. E. Lebani, B. Magnini, N. Novielli (Eds.), Proceedings of the Ninth Italian Conference on Computational Ling… find
  16. A. Zugarini, K. Zeinalipour, S. S. Kadali, M. Maggini, M. Gori, L. Rigutini, Clue-instruct: Text-based clue generation for educational crossword puzzles, in: N. Calzolari, M.-Y. Kan, V. Hoste, A. Lenci, S. Sakti, N. Xue (Eds.), Proceedings of the 2024 Joint International Conference on Computational… find
  17. G. Gallipoli, L. Cagliero, It is not a piece of cake for GPT: Explaining textual entailment recognition in the presence of figurative language, in: O. Rambow, L. Wanner, M. Apidianaki, H. Al-Khalifa, B. D. Eugenio, S. Schockaert (Eds.), Proceedings of the 31st International Conference on Computation…
  18. G. Gallipoli, M. L. Quatra, D. R. Cambrin, S. Greco, L. Cagliero, DANTE at geolingit: Dialect-aware multi-granularity pre-training for locating tweets within italy, in: Proceedings of the Eighth Evaluation Campaign of Natural Language Processing and Speech Tools for Italian. Final Workshop (EVALITA… find
  19. K. Zeinalipour, T. Iaquinta, A. Zanollo, G. Angelini, L. Rigutini, M. Maggini, M. Gori, Italian crossword generator: Enhancing education through interactive word puzzles (2023). find
  20. C. Savelli, F. Giobergia, Enhancing cross-lingual word embeddings: Aligned subword vectors for out-of-vocabulary terms in fasttext, in: 2024 IEEE 18th International Conference on Application of Information and Communication Technologies (AICT), IEEE, 2024, pp. 1–6. find
  21. D. Rege Cambrin, G. Gallipoli, I. Benedetto, L. Cagliero, P. Garza, Beyond accuracy optimization: Computer vision losses for large language model fine-tuning, in: Y. Al-Onaizan, M. Bansal, Y.-N. Chen (Eds.), Findings of the Association for Computational Linguistics: EMNLP 2024, Association for Compu… find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/yassine-2026-ac-dg-cruciverb-it-retrieval/