Elamur Research⌕
☰
home / papers / 2026 / Automatic Classification of Arabic Literature into Historical Eras

Automatic Classification of Arabic Literature into Historical Eras

Paper · 2026 · Zainab Alhathloul and Irfan Ahmad

Summary

The paper asks how well Arabic literary texts can be dated automatically, by assigning them to a historical era. Earlier Arabic work looked almost only at poetry. Here the authors cover both prose and verse. They build two datasets from public corpora: OpenITI for prose and APCD for poetry. Together these run from pre-Islamic times to the present. Era labels follow the six-era scheme of Ghoniem (2017). The authors also test custom time bins, with setups ranging from two classes up to fifteen.

They train fully connected ANNs and RNNs with two tokenization schemes. Each model is run under two split regimes:

  • authors shared between training and test, and
  • author-disjoint splits.

Binary classification works fairly well. Fine-grained setups drop close to the floor. Most errors fall between neighbouring eras, whose boundaries are fuzzy. Poetry turns out easier to date than prose.

The study gives a reproducible baseline for temporal text classification in a language with about 1,500 years of written history. It also shows where the task becomes genuinely hard.

Key points

  • Data: OpenITI (literary prose) and APCD (poetry), both spanning pre-Islamic to modern Arabic. Periodization follows Ghoniem (2017): pre-Islamic, Islamic, Abbasid, Aldoul wa al-emarat, Ottoman, modern.
  • Binary-era F1 is 0.83 on OpenITI and 0.79 on APCD. Fine-grained tasks fall to 0.20 F1 (15 eras, OpenITI) and 0.18 F1 (12 eras, APCD). Five-class OpenITI accuracy ranges from 0.40 to 0.66.
  • ANNs win in most setups. On the binary task, the RNN reaches about 0.91 test accuracy, matching or beating the CNN of Orabi et al. (2020).
  • Preprocessing effects:
    • Removing stop words did not help.
    • Lemmatization helped only for poetry, which has a sparser, richer vocabulary.
  • Author effects: sharing authors across splits boosts prose results, so authorial style leaks into the era signal. For poetry, the split regime barely matters.
  • Linguist-defined eras beat custom time bins when authors are shared across splits. Texts from the Ottoman, Abbasid and Aldoul wa al-emarat periods are identified more readily than Islamic and Modern ones.

Interesting ideas

  • I1. Running every experiment twice, once with authors shared and once author-disjoint, separates "the model learned the period" from "the model learned the writer." This is a cheap control against identity leakage.
  • I2. Lemmatizing paid off only where vocabulary was sparse (poetry), and normalization is not a universal win. The useful lesson is to measure its effect per genre instead of assuming it.
  • I3. Errors clustered on adjacent eras, so the labels are ordinal with soft boundaries rather than independent categories. Reading confusion matrices with that in mind gives a fairer picture than flat accuracy.
  • I4. The authors warn that pretrained transformers trained mostly on modern text may carry a temporal bias against older registers. Any pretrained component can carry such a skew.
  • I5. Proposed future work: extract person names with NER and link them to biographical records to anchor a text in time. This brings external knowledge in instead of relying on surface wording.

For Mova

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

  • I1 (learn): Split evaluation sets by source document or author. This would show whether a parser or reader generalizes or only memorizes one writer's habits.
  • I2 (tokens): Score lemma-based features separately on literary and plain text. Enable aggressive normalization only where sparsity actually hurts downstream steps.
  • I3 (world): When ordering events or assigning time spans, treat neighbouring-period confusions as small errors. Store fuzzy period boundaries in the ontology rather than hard cut-offs.
  • I4 (learn): When a transformer parser or LLM teacher produces silver data, check its agreement with the other witnesses on older or archaic English. Do this before trusting its labels there.
  • I5 (world): Link named people in a text to Wikidata birth and death dates. This would constrain the text's time frame and support explainable reasoning about event order.

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

🎓King Fahd University of Petroleum and MineralsDhahran · Saudi Arabia52 works · 4,666 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 1Research subjects and languages the work is about.Semitic languages · 2,908 works

Cite

Cite as Zainab Alhathloul and Irfan Ahmad. 2026. Automatic Classification of Arabic Literature into Historical Eras. arXiv:2601.16138.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.48550/arxiv.2601.16138 · arXiv 2601.16138

@misc{alhathloul2026automatic,
  title = {{Automatic Classification of Arabic Literature into Historical Eras}},
  author = {Zainab Alhathloul and Irfan Ahmad},
  year = {2026},
  doi = {10.48550/arxiv.2601.16138},
  eprint = {2601.16138},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.48550/arxiv.2601.16138}
}

References

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

  1. Abbas, M., Lichouri, M., & Zeggada, A. (2019). Classification of arabic poems: from the 5th to the 15th century. In M. Cristani, A. Prati, O. Lanz, S. Messelodi, & N. Sebe (Eds.), New trends in image analysis and processing – iciap 2019 (pp. 179–186). Cham: Springer International Publishing. find
  2. Adam, K., Al-Maadeed, S., & Akbari, Y. (2022). Hierarchical fusion using subsets of multi-features for historical arabic manuscript dating. Journal of Imaging, 8(3). Retrieved from https://www.mdpi.com/2313-433X/8/3/60 doi: 10.3390/jimaging8030060 ↗
  3. Al-Fakhoury, H. (1951). History of arabic literature. DMC. find
  4. Alhawarat, M., Hegazi, M., & Hilal, A. (2015). Processing the text of the holy quran: a text mining study. International Journal of Advanced Computer Science and Applications, 6(2), 262–267. ↗
  5. Alrabiah, M., Al-Salman, A., Atwell, E., & Alhelewh, N. (2014). Ksucca a key to exploring arabic historical linguistics. International Journal of Computational Linguistics (IJCL), 5(2), 27–36. find
  6. Belinkov, Y., Magidow, A., Barrón-Cedeño, A., Shmidman, A., & Romanov, M. (2019). Studying the history of the arabic language: language technology and a large-scale historical corpus. Language Resources and Evaluation, 53(4), 771–805. ↗
  7. Chollet, F. (2018). Deep learning with python. Manning. find
  8. Colgrove, C., Chang, S., & Watanaprakornkul, P. Y. (2010). Literary period classification. find
  9. Daif, A. S. (1982). ú ÎëAm.Ì '@ Q唪Ë@ . Islamic Books. find
  10. Darwish, K., & Mubarak, H. (2016, May). Farasa: A new fast and accurate Arabic word segmenter. In N. Calzolari et al. (Eds.), Proceedings of the tenth international conference on language resources and evaluation (LREC’16) (pp. 1070–1074). Portorož, Slovenia: European Language Resources Association (ELRA).
  11. Dietterich, T. G. (1998). Approximate statistical tests for comparing supervised classification learning algorithms. Neural computation, 10(7), 1895–1923. ↗
  12. Elnagar, A., Al-Debsi, R., & Einea, O. (2020). Arabic text classification using deep learning models. Information Processing & Management, 57(1), 102121. ↗
  13. Farha, I. A., & Magdy, W. (2019). Mazajak: An online arabic sentiment analyser. In Proceedings of the fourth arabic natural language processing workshop (pp. 192–198). Association for Computational Linguistics. doi: 10.18653/v1/W19-4621   YmÌ '@ ú G QªË@ H. X B@ . Hebron University Research Journal-B Ghoniem, K. (2017). iÊ¢’ÖÏ @ t' PAJË@ ú ¯ éƒ@P X :Qå•AªÖÏ @ð IK   .  (Humanities)- éJ KA‚ B@ ÐñʪË@) I . KñjJ.ÊË ÉJ Ê mÌ '@ éªÓAg. éÊm.×) , 12(1), 18–34. ↗
  14. Jong, F., Rode, H., & Hiemstra, D. (2005). Temporal language models for the disclosure of historical text. In Humanities, computers and cultural heritage: Proceedings of the xvith international conference of the association for history and computing (ahc 2005) (pp. 161–168). find
  15. Kumar, A., Lease, M., & Baldridge, J. (2011). Supervised language modeling for temporal resolution of texts. In Proceedings of the 20th acm international conference on information and knowledge management (pp. 2069–2072). find
  16. Lee, J., Cho, K., & Hofmann, T. (2017). Fully character-level neural machine translation without explicit segmentation. Transactions of the Association for Computational Linguistics, 5, 365–378. ↗
  17. McMahon, A. M. (1994). Understanding language change. Cambridge, UK: Cambridge university press. ↗
  18. Mihalcea, R., & Nastase, V. (2012). Word epoch disambiguation: Finding how words change over time. In Proceedings of the 50th annual meeting of the association for computational linguistics (volume 2: Short papers) (Vol. 2, pp. 259–263). find
  19. Niculae, V., Zampieri, M., Dinu, L. P., & Ciobanu, A. M. (2014). Temporal text ranking and automatic dating of texts. In Proceedings of the 14th conference of the european chapter of the association for computational linguistics, volume 2: Short papers (pp. 17–21). find
  20. Orabi, M., El Rifai, H., & Elnagar, A. (2020). Classical arabic poetry: Classification based on era. In Proceedings of ieee/acs international conference on computer systems and applications, aiccsa (pp. 1–6). ↗
  21. Özyurt, F., Sert, E., & Avcı, D. (2020). An expert system for brain tumor detection: Fuzzy c-means with super resolution and convolutional neural network with extreme learning machine. Medical hypotheses, 134, 109433. ↗
  22. Popescu, O., & Strapparava, C. (2015). Semeval 2015, task 7: Diachronic text evaluation. In Proceedings of the 9th international workshop on semantic evaluation (semeval 2015) (pp. 870–878). find
  23. Porat, E., Koppel, M., & Shmidman, A. (2018). Identification of parallel passages across a large hebrew/aramaic corpus. Journal of Data Mining & Digital Humanities, Special Issue on Computer-Aided Processing of Intertextuality in Ancient Languages. doi: 10.46298/jdmdh.1388 ↗
  24. Romanov, M., & Seydi, M. (2019). Openiti: a machine-readable corpus of islamicate texts. Zenodo, URL: https://doi.org/10.5281/zenodo, 3082464. ↗
  25. Ruma, J. F., Akter, S., Laboni, J. J., & Rahman, R. M. (2022). A deep learning classification model for persian hafez poetry based on the poet’s era. Decision Analytics Journal, 4, 100111. Retrieved from https://www.sciencedirect.com/science/article/pii/S2772662222000479 doi: https://doi.org/10.1016/j.dajour.2022.100111 ↗
  26. Salaberri, H., Salaberri, I., Arregi, O., & Zapirain, B. (2015). Ixagroupehudiac: A multiple approach system towards the diachronic evaluation of texts. In Proceedings of the 9th international workshop on semantic evaluation (semeval 2015) (pp. 840–845). find
  27. Štajner, S., & Zampieri, M. (2013). Stylistic changes for temporal text classification. In International conference on text, speech and dialogue (pp. 519–526). find
  28. Szymanski, T., & Lynch, G. (2015). Ucd: Diachronic text classification with character, word, and syntactic n-grams. In Proceedings of the 9th international workshop on semantic evaluation (semeval 2015). united states. find
  29. Tilahun, G., Feuerverger, A., Gervers, M., et al. (2012). Dating medieval english charters. The Annals of applied statistics, 6(4), 1615–1640. ↗
  30. Uysal, A. K., & Gunal, S. (2014). The impact of preprocessing on text classification. Information processing & management, 50(1), 104–112. ↗
  31. Yousef, W. A., Ibrahime, O. M., Madbouly, T. M., Mahmoud, M. A., El-Kassas, A. H., Hassan, A. O., & Albohy, A. R. (2018, July). Poem comprehensive dataset (pcd). Retrieved from https://hci-lab.github.io/ArabicPoetry -1-Private/#PCD Zampieri, M., Ciobanu, A. M., Niculae, V., & Dinu, L. P. (2015). Ambra: A ranking approach to temporal text classification. In Proceedings of the 9th international workshop on semantic evaluation (semeval 2015) (pp. 851–855). Association for Computational Linguistics. ↗

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/alhathloul-2026-automatic-classification-arabic-literature-historical/