Elamur Research⌕
☰
home / papers / 2026 / Tarikhi: Arabic Temporal Information Extraction from Arabic Historical Documents

Tarikhi: Arabic Temporal Information Extraction from Arabic Historical Documents

Paper · 2026 · Qusay Abdo, Serin Atiani, Tariq Sraiji et al.

Summary

Arabic historical books hold many accounts of political and social events, but few Arabic tools can turn them into structured data. Scanned books add further problems: OCR errors, inconsistent spelling and long, winding narratives. The authors built Tarikhi, an automated retrieval-augmented generation pipeline that turns scanned Arabic books into event records with a description, date and location.

The pipeline has these stages:

  • OCR the pages and clean the text.
  • Split the text into overlapping, token-based chunks.
  • Find candidate event spans with Arabic NER.
  • Sharpen those spans with the Silma-9B LLM.
  • Retrieve supporting chunks to extract each event's attributes.
  • Merge duplicates into canonical events and link them with SAME_AS and RELATED_TO relations.

The authors tested it on chapters from two modern books about Middle Eastern political history. Two researchers checked the output and accepted 75 of 95 canonical events, giving an F1 of 75.3%. The work shows that end-to-end temporal event extraction from long Arabic texts is feasible, which could make undigitized historical narratives searchable for digital humanities.

Key points

  • OCR choice: Surya OCR, DeepSeek OCR, Qari and a majority-vote hybrid were compared on 20 hand-transcribed pages, 10 from each book. Surya scored best, with 0.8592 character accuracy and 0.9366 word accuracy, measured with difflib.SequenceMatcher. The hybrid scored 0.7746 and 0.8829, so it did not beat Surya alone.
  • Chunking and retrieval: chunks are 256 tokens with a 50-token overlap, cut with the silma-9b tokenizer. Retrieval embeddings come from intfloat/multilingual-E5-large.
  • Sources: the Arabic translation of Rashid Khalidi's The Hundred Years' War on Palestine (2021) and Amin Howeidi's The 1967 War: Secrets and Mysteries (2006).
  • Results: 156 raw events were reduced to 95 canonical ones, of which 75 were judged correct. Reported scores are 78.8% recall, 72.1% precision and 75.3% F1.
  • Limitations: only two date-heavy books were used, the judging is subjective, and individual components were not evaluated, because no long-form Arabic gold corpora exist.

Interesting ideas

  • I1. Event spans come from a trained NER model, and an LLM then rewrites them using the surrounding context, which pairs reproducible anchors with richer, more descriptive names.
  • I2. Duplicate events are merged into canonical records using both semantic and temporal similarity, then typed as SAME_AS or RELATED_TO links.
  • I3. A cheap hand-made gold set of 20 pages showed that a voting ensemble scored below the best single system.
  • I4. Errors were sorted into concrete types: non-events such as person or organization names, correct events with a wrong date, and correct events with a wrong location.
  • I5. Long documents are chunked with the downstream model's own tokenizer and with overlap, so no event is cut at a boundary.

For Mova

How these ideas could serve Mova, our small open language model — its files, stages and tests:

  • I1 (frames): the UD-based frame extractor could fix the predicate span, and an optional refiner could only improve labels inside that span, so the explainable anchor stays intact.
  • I2 (world): an event-coreference step could merge mentions that share a predicate class, participants and time anchor, and record explicit identity versus relatedness edges.
  • I3 (learn): witness-parser agreement for silver data should be checked against a small gold sample, not assumed to beat the strongest witness.
  • I4 (answer): QA evaluation on tales could tag failures by type (wrong entity, wrong time, wrong place) to show which stage broke.
  • I5 (world): world-building over long texts could use overlapping windows and reconcile entities across the overlap zones.

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

🎓Princess Sumaya University for TechnologyAmman · Jordan6 works · 63 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 worksLiterature and Literary Theory · 519 works

🧭 Topics 2Research subjects and languages the work is about.Semitic languages · 2,908 worksInformation extraction · 167 works

Cite

Cite as Qusay Abdo, Serin Atiani, Tariq Sraiji and Adnan Saeed. 2026. Tarikhi: Arabic Temporal Information Extraction from Arabic Historical Documents. Proceedings of the 2nd International Workshop on Nakba Narratives as Language Resources @ LREC 2026.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.63317/5eco9hgcoxp8

@inproceedings{abdo2026tarikhi,
  title = {{Tarikhi: Arabic Temporal Information Extraction from Arabic Historical Documents}},
  author = {Qusay Abdo, Serin Atiani, Tariq Sraiji and Adnan Saeed},
  booktitle = {Proceedings of the 2nd International Workshop on Nakba Narratives as Language Resources @ LREC 2026},
  year = {2026},
  doi = {10.63317/5eco9hgcoxp8},
  url = {https://doi.org/10.63317/5eco9hgcoxp8}
}

References

39 references, 9 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. SILMA AI. 2024. Silma-9b-instruct-v1.0. https://huggingface.co/silma-ai/SILMA-9B-Instruct-v1.0. Large Language Model. ↗
  2. Abdulaziz M. Alayba. 2025. Arabic natural language processing (nlp): A comprehensive review of challenges, techniques, and emerging trends. Computers, 14(11). ↗
  3. Alaa Aljabari, Lina Duaibes, Mustafa Jarrar, and Mohammed Khalilia. 2024a. Event-arguments extraction corpus and modeling using bert for arabic.
  4. Alaa Aljabari, Lina Duaibes, Mustafa Jarrar, and Mohammed Khalilia. 2024b. Event-arguments extraction corpus and modeling using bert for arabic. arXiv preprint arXiv:2407.21153. ↗
  5. Alaa Aljabari, Mohammed Khalilia, and Mustafa Jarrar. 2025. Wojood Relations : Arabic relation extraction corpus and modeling. In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing, pages 34342– 34360. ﺃ ،ﻱ Amin Huwaidi. 2006. Harb 1967: Asrar wa Khabaya , ﻭ ﺭﺍ ﺃ : 1967 ﺏ . Al-Maktab Al-Misri Al-Hadith, ﺍ ﻱ ﺍ ﺍ, Cairo, Egypt. find
  6. Wissam Antoun, Fady Baly, and Hazem Hajj. 2020. Arabert: Transformer-based model for arabic language understanding. In Proceedings of the 4th Workshop on Open-Source Arabic Corpora and Processing Tools, with a Shared Task on Offensive Language Detection (OSACT 2020), pages 9–15. OSACT 2020. ↗
  7. M. H. Bashir, A. M. Azmi, H. Nawaz, et al. 2023. Arabic natural language processing for qur’anic research: A systematic review. Artificial Intelligence Review, 56:6801–6854. ↗
  8. Steven Bethard, Leon Derczynski, Guergana Savova, James Pustejovsky, and Marc Verhagen. 2015. SemEval-2015 task 6: Clinical TempEval. In Proceedings of the 9th International Workshop on Semantic Evaluation (SemEval 2015), pages 806–814. find
  9. Tarik Boudaa, Mohamed El Marouani, and Nourddine Enneya. 2018. Arabic temporal expression tagging and normalization. In Big Data, Cloud and Applications (BDCA 2018), volume 872 of Communications in Computer and Information Science, pages 271–284. Springer. find
  10. Bibo Cai, Xiao Ding, Zhouhao Sun, Bing Qin, Ting Liu, Baojun Wang, and Lifeng Shang. 2023. Self-supervised logic induction for explainable fuzzy temporal commonsense reasoning. In Proceedings of the Thirty-Seventh AAAI Conference on Artificial Intelligence and Thirty-Fifth Conference on Innovative Applications of Artificial Intelligence and Thirteenth Symposium on Educational Advances in Artificial Intelligence, AAAI’23/IAAI’23/EAAI’23. AAAI Press. find
  11. Surya OCR Contributors. 2024. Surya ocr: Multilingual document ocr toolkit. https://Open-github.com/datalab-to/surya. source OCR toolkit supporting 90+ languages. ↗
  12. Kareem Darwish, Nizar Habash, Mourad Abbas, Hend Al-Khalifa, Huseein T. Al-Natsheh, Houda Bouamor, Karim Bouzoubaa, Violetta Cavalli-Sforza, Samhaa R. El-Beltagy, Wassim El-Hajj, Mustafa Jarrar, and Hamdy Mubarak. 2021. A panoramic survey of natural language processing in the arab world. Commun. ACM, 64(4):72– 81. ↗
  13. Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. Bert: Pre-training of deep bidirectional transformers for language understanding. In North American Chapter of the Association for Computational Linguistics.
  14. Jennifer D’Souza and Vincent Ng. 2014. Annotating inter-sentence temporal relations in clinical notes. In Proceedings of the Ninth International Conference on Language Resources and Evaluation (LREC’14), pages 2758–2765. find
  15. Michele Filannino and Goran Nenadić. 2015. Temporal expression extraction with extensive feature type selection and a posteriori label adjustment. Data & Knowledge Engineering, 100:19– 33. ↗
  16. Abbas Ghaddar, Yimeng Wu, Ahmad Rashid, Khalil Bibi, Mehdi Rezagholizadeh, Chao Xing, Yasheng Wang, Duan Xinyu, Zhefeng Wang, Baoxing Huai, Xin Jiang, Qun Liu, and Philippe Langlais. 2022. Jaber and saber: Junior and senior arabic bert. arXiv preprint arXiv:2112.04329. ↗
  17. Nafaa Haffar, Emna Hkiri, and Mounir Zrigui. 2020. Enrichment of arabic timeml corpus. In Computational Collective Intelligence (ICCCI 2020), volume 12496 of Lecture Notes in Computer Science, pages 233–244. Springer. find
  18. Tymaa Hammouda, Mustafa Jarrar, and Mohammed Khalilia. 2024. Sinatools: Open source toolkit for arabic natural language processing. Procedia Computer Science, 244:388–396. 6th International Conference on AI in Computational Linguistics.
  19. Kuan-Hao Huang, I-Hung Hsu, Tanmay Parekh, Zhiyu Xie, Zixuan Zhang, Premkumar Natarajan, Kai-Wei Chang, Nanyun Peng, and Heng Ji. 2024. Textee: Benchmark, reevaluation, reflections, and future challenges in event extraction. find
  20. Mustafa Jarrar, Mohammed Khalilia, and Sana Ghanem. 2022. Wojood: Nested arabic named entity corpus and recognition using bert. In Proceedings of the Thirteenth Language Resources and Evaluation Conference (LREC 2022), pages 3626–3636.
  21. Rashid Khalidi. 2021. ��� ����� ��� ��� ������. ��� ,������ .����� Arabic translation of ”The Hundred Years’ War on Palestine”. find
  22. Panteleimon Krasadakis, Evangelos Sakkopoulos, and Vassilios S. Verykios. 2024. A survey on challenges and advances in natural language processing with a focus on legal informatics and low-resource languages. Electronics, 13(3). ↗
  23. Artuur Leeuwenberg and Marie-Francine Moens. 2018. Temporal information extraction by predicting relative time-lines. In Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, pages 1237–1246. find
  24. Chahira Lhioui, Anis Zouaghi, and Mounir Zrigui. 2017. A rule-based approach for arabic temporal expression extraction. In 2017 International Conference on Engineering MIS (ICEMIS), pages 1–6. find
  25. Nasrin Mostafazadeh, Alyson Grealish, Nathanael Chambers, James Allen, and Lucy Vanderwende. 2016. CaTeRS: Causal and temporal relation scheme for semantic annotation of event structures. In Proceedings of the Fourth Workshop on Events, pages 51–61. find
  26. Qiang Ning, Sanjay Subramanian, and Dan Roth. 2019. An improved neural baseline for temporal relation extraction. 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 6203–6209. ↗
  27. James Pustejovsky, José M. Castaño, Robert Ingria, Roser Saurí, Robert J. Gaizauskas, Andrea Setzer, Graham Katz, and Dragomir R. Radev. 2003. Timeml: Robust specification of event and temporal expressions in text. In New Directions in Question Answering. find
  28. Iman Saleh, Lamia Tounsi, and Josef van Genabith. 2011. Zaman and raqm: Extracting temporal and numerical expressions in arabic. In Information Retrieval Technology, pages 562– 573. find
  29. Naushad UzZaman, Hector Llorens, Leon Derczynski, James Allen, Marc Verhagen, and James Pustejovsky. 2013. SemEval-2013 task 1: TempEval-3: Evaluating time expressions, events, and temporal relations. In Second Joint Conference on Lexical and Computational Semantics (*SEM), Volume 2: Proceedings of the Seventh International Workshop on Semantic Evaluation (SemEval 2013), pages 1–9. find
  30. Marc Verhagen, Robert Gaizauskas, Frank Schilder, Mark Hepple, Graham Katz, and James Pustejovsky. 2007. SemEval-2007 task 15: TempEval temporal relation identification. find
  31. In Proceedings of the Fourth International Workshop on Semantic Evaluations (SemEval-2007), pages 75–80. find
  32. Liang Wang, Nan Yang, Xiaolong Huang, Linjun Yang, Rangan Majumder, and Furu Wei. 2024. Multilingual e5 text embeddings: A technical report. ArXiv:2402.05672v1 [cs.CL, cs.IR], Technical Report, submitted 2024-02-08. ↗
  33. Ahmed Wasfy, Omer Nacar, Abdelakreem Elkhateb, Mahmoud Reda, Omar Elshehy, Adel Ammar, and Wadii Boulila. 2025. Qari-ocr: High-fidelity arabic text recognition through multimodal large language model adaptation. arXiv preprint arXiv:2506.02295. ↗
  34. Haoran Wei, Yaofeng Sun, and Yukun Li. 2025. Deepseek-ocr: Contexts optical compression. arXiv preprint arXiv:2510.18234. ↗
  35. Georg Wenzel and Adam Jatowt. 2023. An overview of temporal commonsense rea-arXiv preprint soning and acquisition. arXiv:2307.15394. ↗
  36. Kam-Fai Wong, Yunqing Xia, Wenjie Li, and Chunfa Yuan. 2005. An overview of temporal information extraction. International Journal of Computer Processing of Languages, 18(2):219– 240. ↗
  37. Fadi A. Zaraket. 2012. Arabic temporal entity extraction using morphological analysis. find
  38. Xiaobin Zhang, Liangjun Zang, Qianwen Liu, Shuchong Wei, and Songlin Hu. 2024. Event temporal relation extraction based on retrieval-augmented on llms. find
  39. Y. Zhang and Q. Han. 2025. Enhancing pre-trained language model by answering natural questions for event extraction. Frontiers in Artificial Intelligence, 8:1520290. ↗

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/abdo-2026-tarikhi-arabic-temporal-information-extraction/