--- title: "From Arabic Text to Puzzles — paper by Zeinalipour et al. (2025)" description: "It presents a system that turns any Arabic educational text into crossword material: answer keywords… 3 references, cited by 3 papers here." authors: ["Kamyar Zeinalipour", "Moahmmad Saad", "Marco Maggini", "Marco Gori"] year: 2025 venue: "Proceedings of the First Workshop on Language Models for Low-Resource Languages" source: "https://aclanthology.org/2025.loreslm-1.36/" licence: "CC BY 4.0 (ACL Anthology)" licence_url: "https://creativecommons.org/licenses/by/4.0/" summary_licence: "Elamur Cosmic License" canonical: "https://research.elamur.ai/free/en/papers/zeinalipour-2025-arabic-text-puzzles-llm-driven/" --- [Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/zeinalipour-2025-arabic-text-puzzles-llm-driven/) [home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2025](https://md.research.elamur.ai/free/en/papers/2025/page/12/) / From Arabic Text to Puzzles: LLM-Driven Development of Arabic Educational Crosswords # From Arabic Text to Puzzles: LLM-Driven Development of Arabic Educational Crosswords **Authors:** [Kamyar Zeinalipour](https://md.research.elamur.ai/free/en/authors/kamyar-zeinalipour/), Moahmmad Saad, [Marco Maggini](https://md.research.elamur.ai/free/en/authors/marco-maggini/) and [Marco Gori](https://md.research.elamur.ai/free/en/authors/marco-gori/) · **Year:** 2025 · **Venue:** [Proceedings of the First Workshop on Language Models for Low-Resource Languages](https://md.research.elamur.ai/free/en/venues/loreslm/) ([Workshop on Language Models for Low-Resource Languages 2025](https://md.research.elamur.ai/free/en/venues/loreslm/2025/)) · **Publisher:** [Association for Computational Linguistics](https://md.research.elamur.ai/free/en/publishers/association-for-computational-linguistics/) · **Licence:** [CC BY 4.0 (ACL Anthology)](https://creativecommons.org/licenses/by/4.0/) **Links:** [Publisher's page](https://aclanthology.org/2025.loreslm-1.36/) · [PDF](https://aclanthology.org/2025.loreslm-1.36.pdf) · [OpenAlex](https://openalex.org/W4406733726) **Citations:** 0 citations (OpenAlex, 2026-10-14) · 3 in our library, of the 149,439 papers whose bibliographies we read ## Summary The paper addresses the lack of Arabic educational tools. It presents a system that turns any Arabic educational text into crossword material: answer keywords plus matching clues. The authors built Arabic-Clue-Instruct from Arabic Wikipedia. They took each article's opening section, used its bolded terms as candidate answers, and filtered articles and keywords by popularity, relevance and length. GPT-4-Turbo then wrote several context-grounded clues per entry, guided by a carefully engineered prompt in the spirit of SELF-INSTRUCT. This data was used to fine-tune GPT-3.5-Turbo and Llama3-8B-Instruct. ROUGE overlap with GPT-4-Turbo clues rose after fine-tuning. Human judges, using a five-grade scale, saw a much larger effect: the fine-tuned 8B Llama produced top-grade clues far more often than its base version. Teachers can then pick from the generated clues to assemble puzzles. The dataset and the fine-tuned Llama model are openly released. This gives a low-resource language a reusable resource and a working recipe for educational gamification. ## Key points - Roughly 211,000 Arabic Wikipedia articles were filtered down to 14,497 entries across 20 thematic categories; of these, 14,000 went to the training set and the remaining 497 were held out for testing. - GPT-4-Turbo generated at least three clues per entry, for 54,196 unique clues in Arabic-Clue-Instruct. - Filtering rules: - articles under 50 words were dropped; - keywords longer than two words were removed; - keywords had to fall within a 3–20 character range. - Mean ROUGE-1 against GPT-4-Turbo clues: - GPT-3.5-Turbo: 0.0152 base, 0.0405 fine-tuned; - Llama3-8B-Instruct: 0.0063 base, 0.0354 fine-tuned. - Human evaluation covered 200 contexts with 3 clues each. The share of grade "A" clues changed as follows: - Llama3-8B-Instruct: 36.02% base, 78.86% fine-tuned; - GPT-3.5-Turbo: 28.47% base, 54.33% fine-tuned. ## Interesting ideas - Bolded terms in a Wikipedia lead section serve as free, structure-derived labels for an article's central concepts, so answer keywords need no manual annotation. - A strong proprietary LLM produced the training data, and a small open 8B model fine-tuned on it more than doubled its top-grade output. A modest local model can therefore inherit a narrow skill from a large teacher. - The authors reported ROUGE but explicitly distrusted it, since even good clues scored near zero. They relied on a graded human judgment instead, which shows surface overlap is a poor proxy for whether a generated text is valid. - Clues were built by weaving facts from the surrounding passage around a target keyword, so a clue amounts to a compact, context-grounded definition whose answer is a single term. - The pipeline over-generates candidate clues and leaves the final choice to educators. This human-in-the-loop design keeps automation useful even when individual outputs are imperfect. ## For Mova How these ideas could serve Mova, our small open language model — its files, stages and tests: - **I1** (world): the bolded lead term could be taken as the canonical name of an article's main entity when linking Wikidata items into the knowledge graph, giving cheap anchor labels. - **I2** (learn): this supports an LLM teacher tier that writes task-specific examples once, after which a small local component is trained and checked on held-out items. - **I3** (learn): evaluations of generated or rephrased output should use meaning-level checks or graded human/LLM judgments, with surface metrics only as a secondary signal. - **I4** (back): the world graph could drive clue-style generation, describing an entity through its stored states and relations without naming it, which tests whether the graph holds enough meaning. - **I5** (answer): school-course test items generated from a text could be produced in surplus, each with its explainable source path, for a teacher or checker to select from. *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 - [University of Siena](https://md.research.elamur.ai/free/en/universities/university-of-siena/) — University, Siena, Italy · 9 works · 198 citations 📚 Shelves (7): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 16,667 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 16,194 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 16,039 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 15,891 works · [🤝 Social Sciences](https://md.research.elamur.ai/free/en/shelves/social-sciences/) · 1,772 works · [🗄️ Information Systems](https://md.research.elamur.ai/free/en/shelves/computer-science/information-systems/) · 1,437 works · [🎓 Education](https://md.research.elamur.ai/free/en/shelves/social-sciences/education/) · 203 works 🧭 Topics (2): [🧠 Language models & pretraining](https://md.research.elamur.ai/free/en/topics/language-models/) · 3,320 works · [Semitic languages](https://md.research.elamur.ai/free/en/topics/semitic-languages/) · 2,908 works ## Cite Cite as: Kamyar Zeinalipour, Moahmmad Saad, Marco Maggini and Marco Gori. 2025. From Arabic Text to Puzzles: LLM-Driven Development of Arabic Educational Crosswords. Proceedings of the First Workshop on Language Models for Low-Resource Languages. The paper itself, not this page. BibTeX from the library's record of the work. ```bibtex @inproceedings{zeinalipour2025arabic, title = {{From Arabic Text to Puzzles: LLM-Driven Development of Arabic Educational Crosswords}}, author = {Kamyar Zeinalipour, Moahmmad Saad, Marco Maggini and Marco Gori}, booktitle = {Proceedings of the First Workshop on Language Models for Low-Resource Languages}, year = {2025}, url = {https://aclanthology.org/2025.loreslm-1.36/} } ``` - [← previous FinNLI: Novel Dataset for Multi-Genre Financial Natural Language Inference Benchmarking](https://md.research.elamur.ai/free/en/papers/magomere-2025-finnli-novel-dataset-multi-genre/) - [next → From Disagreement to Understanding: The Case for Ambiguity Detection in NLI](https://md.research.elamur.ai/free/en/papers/jayaweera-2025-disagreement-understanding-case-ambiguity-detection/) - [↑ back to papers of 2025](https://md.research.elamur.ai/free/en/papers/2025/page/12/) ## References 3 references, 1 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet. 1. Bhavna Arora and NS Kumar. 2019. Automatic keyword extraction and crossword generation tool for indian languages: Seekh. In 2019 IEEE Tenth International Conference on Technology for Education (T4E), pages 272–273. IEEE. [find](https://scholar.google.com/scholar?q=Bhavna+Arora+and+NS+Kumar%2E+2019%2E+Automatic+keyword+extraction+and+crossword+generation+tool+for+indian+languages%3A+Seekh%2E+In+2019+IEEE+Tenth+International+Conference+on+Technology+for+Education+%28T4E%29%2C+) 2. Yolanda Dita Bella and Endang Mastuti Rahayu. 2023. The improving of the student’s vocabulary achievement through crossword game in the new normal era. Edunesia: Jurnal Ilmiah Pendidikan, 4(2):830–842. [↗](https://doi.org/10.51276/edu.v4i2.433) 3. Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, et al. 2020. [Language models are few-shot learners](https://md.research.elamur.ai/free/en/papers/brown-2020-language-models-few-shot-learners/). Advances in neural information processing systems, 33:1877–1901. *Updated 1 November 2026*