---
title: "Automated Authentication of Quranic Verses Using BERT"
description: "The work matters because it builds a negative class of hard near-misses, a harder setting than earlier studies… 99.75% recall. 24 references."
authors: ["Khubaib Amjad Alam", "Maryam Khalid", "Syed Ahmed Ali", "Haroon Mahmood", "Qaisar Shafi", "Muhammad Haroon", "Zulqarnain Haider"]
year: 2025
venue: "Proceedings of the New Horizons in Computational Linguistics for Religious Texts"
source: "https://aclanthology.org/2025.clrel-1.6/"
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/alam-2025-automated-authentication-quranic-verses-bert/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/alam-2025-automated-authentication-quranic-verses-bert/)

[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/17/) / Automated Authentication of Quranic Verses Using BERT (Bidirectional Encoder Representations from Transformers) based Language Models

# Automated Authentication of Quranic Verses Using BERT (Bidirectional Encoder Representations from Transformers) based Language Models

**Authors:** Khubaib Amjad Alam, Maryam Khalid, Syed Ahmed Ali, Haroon Mahmood, Qaisar Shafi, Muhammad Haroon and Zulqarnain Haider · **Year:** 2025 · **Venue:** [Proceedings of the New Horizons in Computational Linguistics for Religious Texts](https://md.research.elamur.ai/free/en/venues/clrel/) ([New Horizons in Computational Linguistics for Religious Texts 2025](https://md.research.elamur.ai/free/en/venues/clrel/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.clrel-1.6/) · [PDF](https://aclanthology.org/2025.clrel-1.6.pdf)

## Summary
The paper addresses checking whether a Quranic verse circulating online is genuine. The authors frame this as binary classification: an authentic verse versus a slightly altered imitation. Exact string matching or hashing can confirm identical copies, but it says little about near-identical edits that shift meaning. Authentic verses come from tanzil.net. The non-authentic class started with a manually corrupted seed and was then enlarged automatically. The generator modified important words, using TF-IDF to pick them and WordNet antonyms to replace them, and deleted words when no antonym was found. Sentence-BERT cosine similarity kept each corrupted verse close to its source. Three Arabic-pretrained encoders were fine-tuned on this data: BERT-Base-Arabic, AraBERT and MarBERT. MarBERT scored best with an F1 of 93.80%, BERT-Base-Arabic was close behind, and AraBERT largely failed. The work matters because it builds a negative class of hard near-misses, a harder setting than earlier studies that contrasted Quranic text with ordinary Arabic prose.

## Key points
- The authentic class is taken from tanzil.net. The authors list this single source as a limitation, since it does not represent other Qira'at (canonical recitation variants).
- The non-authentic class began with 500 hand-altered verses. Automated generation and quality checks expanded it to 2068 altered verses, which are kept internal and will not be released.
- Test results (accuracy / precision / recall / F1):
  - MarBERT: 93.73 / 91.25 / 96.50 / 93.80
  - BERT-Base-Arabic: 92.75 / 89.56 / 96.50 / 92.90
- AraBERT reached 99.75% recall but only 53.92% precision, giving 57.99% accuracy and 70.00% F1. It flagged almost everything as one class.
- Planned future work includes:
  - Qira'at-aware data
  - paraphrase and typo augmentation
  - a real-time checking tool
  - transfer to other scriptures via mBERT or XLM-R

## Interesting ideas
- Build hard negatives by making small, targeted edits to genuine text and keeping only those whose sentence-embedding similarity to the original stays above a threshold, so the classifier cannot win on surface cues alone.
- The edit recipe is cheap and layered: pick the most informative word by TF-IDF, swap in its WordNet antonym, and fall back to deleting the word when no antonym exists.
- Start with a small hand-made seed of corruptions, then scale it with automation plus quality checks, rather than relying entirely on either manual work or generation.
- AraBERT's near-perfect recall alongside coin-flip precision shows that a single headline number can hide a model that has collapsed onto one label.

## For Mova

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

- **I1** (learn): Mova could generate near-miss variants of evaluation sentences as negative controls and confirm that its readers' answers change whenever the edit alters the meaning.
- **I2** (world): Applying this to the key verb or adjective of a fable gives a minimal pair, letting Mova test whether the world it builds (states, holdings, outcomes) actually flips.
- **I3** (learn): Silver-data repair rules could follow the same pattern, beginning with a few curated error examples that are then expanded automatically and screened before entering training.
- **I4** (answer): School-test and QA evaluations should report per-class precision and recall, and include a trivial always-same-answer baseline, so that degenerate readers are caught early.

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

📚 Shelves (2): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 18,614 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 17,965 works

🧭 Topics (1): [🧠 Language models & pretraining](https://md.research.elamur.ai/free/en/topics/language-models/) · 3,894 works

## Cite

Cite as: Khubaib Amjad Alam, Maryam Khalid, Syed Ahmed Ali, Haroon Mahmood, Qaisar Shafi, Muhammad Haroon and Zulqarnain Haider. 2025. Automated Authentication of Quranic Verses Using BERT (Bidirectional Encoder Representations from Transformers) based Language Models. Proceedings of the New Horizons in Computational Linguistics for Religious Texts.

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

```bibtex
@inproceedings{alam2025automated,
  title = {{Automated Authentication of Quranic Verses Using BERT (Bidirectional Encoder Representations from Transformers) based Language Models}},
  author = {Khubaib Amjad Alam, Maryam Khalid, Syed Ahmed Ali, Haroon Mahmood, Qaisar Shafi, Muhammad Haroon and Zulqarnain Haider},
  booktitle = {Proceedings of the New Horizons in Computational Linguistics for Religious Texts},
  year = {2025},
  url = {https://aclanthology.org/2025.clrel-1.6/}
}
```

- [← previous AtlasIA at SemEval-2025 Task 11: FastText-Based Emotion Detection in Moroccan Arabic for Low-Resource Settings](https://md.research.elamur.ai/free/en/papers/majjodi-2025-atlasia-semeval-2025-task-11/)
- [next → Automatic Proficiency Assessment in L2 English Learners](https://md.research.elamur.ai/free/en/papers/mohammadi-2025-automatic-proficiency-assessment-l2-english/)
- [↑ back to papers of 2025](https://md.research.elamur.ai/free/en/papers/2025/page/17/)

## References

24 references, 7 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Mustafa Jarrar, Fadi Zaraket, Rami Asia, and Hamzeh Amayreh. 2018. Diacritic-based matching of arabic words. ACM Transactions on Asian and Low-Resource Language Information Processing (TAL-LIP), 18(2):1–21. [↗](https://doi.org/10.1145/3242177)
2. Semantic textual similarity using sbert. Accessed: November 2024. [find](https://scholar.google.com/scholar?q=Semantic+textual+similarity+using+sbert%2E+Accessed%3A+November+2024%2E)
3. Ali Saleh Alammary. 2022. Bert models for arabic text classification: a systematic review. Applied Sciences, 12(11):5720. [↗](https://doi.org/10.3390/app12115720)
4. Amirrudin Kamsin, Abdullah Gani, Ishak Suliaman, Salinah Jaafar, Rohana Mahmud, Aznul Qalid Md Sabri, Zaidi Razak, Mohd Yamani Idna Idris, Maizatul Akmar Ismail, Noorzaily Mohamed Noor, et al. 2014. Developing the novel quran and hadith authentication system. In The 5th International Conference on Information and Communication Technology for The Muslim World (ICT4M), pages 1–5. IEEE. [find](https://scholar.google.com/scholar?q=Amirrudin+Kamsin%2C+Abdullah+Gani%2C+Ishak+Suliaman%2C+Salinah+Jaafar%2C+Rohana+Mahmud%2C+Aznul+Qalid+Md+Sabri%2C+Zaidi+Razak%2C+Mohd+Yamani+Idna+Idris%2C+Maizatul+Akmar+Ismail%2C+Noorzaily+Mohamed+Noor%2C+et+al%2E+2014%2E+D)
5. Mohamed Alkaoud and Mairaj Syed. 2020. [On the importance of tokenization in arabic embedding models](https://md.research.elamur.ai/free/en/papers/alkaoud-2020-importance-tokenization-arabic-embedding-models/). In Proceedings of the fifth Arabic natural language processing workshop, pages 119–129. [↗](https://doi.org/10.48448/1tvq-g003)
6. Badr AlKhamissi, Mohamed Gabr, Muhammad El-Nokrashy, and Khaled Essam. 2021. [Adapting marbert for improved arabic dialect identification: Submission to the nadi 2021 shared task](https://md.research.elamur.ai/free/en/papers/alkhamissi-2021-adapting-marbert-improved-arabic-dialect/). arXiv preprint arXiv:2103.01065. [↗](https://arxiv.org/abs/2103.01065)
7. Mikhail V Koroteev. 2021. Bert: a review of applications in natural language processing and understanding. arXiv preprint arXiv:2103.11943. [↗](https://arxiv.org/abs/2103.11943)
8. Abdullah Y Muaad, Shaina Raza, Usman Naseem, and Hanumanthappa J Jayappa Davanagere. 2023. Arabic text detection: a survey of recent progress challenges and opportunities. Applied Intelligence, 53(24):29845–29862. [↗](https://doi.org/10.1007/s10489-023-04992-9)
9. Wafa Abdullah Alrajhi, Hend Al-Khalifa, and Abdulmalik AlSalman. 2022. [Assessing the linguistic knowledge in arabic pre-trained language models using minimal pairs](https://md.research.elamur.ai/free/en/papers/alrajhi-2022-assessing-linguistic-knowledge-arabic-pre/). In Proceedings of the Seventh Arabic Natural Language Processing Workshop (WANLP), pages 185–193.
10. N Reimers. 2019. [Sentence-bert: Sentence embeddings using siamese bert-networks](https://md.research.elamur.ai/free/en/papers/reimers-2019-sentence-bert-sentence-embeddings-siamese/). arXiv preprint arXiv:1908.10084. [↗](https://arxiv.org/abs/1908.10084)
11. Wissam Antoun, Fady Baly, and Hazem Hajj. 2020. [Arabert: Transformer-based model for arabic language understanding](https://md.research.elamur.ai/free/en/papers/antoun-2020-arabert-transformer-based-model-arabic/). arXiv preprint arXiv:2003.00104. [↗](https://arxiv.org/abs/2003.00104)
12. S Shreyashree, Pramod Sunagar, S Rajarajeswari, and Anita Kanavalli. 2022. A literature review on bidirectional encoder representations from transformers. Inventive Computation and Information Technologies: Proceedings of ICICIT 2021, pages 305–320. [↗](https://doi.org/10.1007/978-981-16-6723-7_23)
13. Arafat Awajan. 2007. Arabic text preprocessing for the natural language processing applications. Arab Gulf Journal of Scientific Research, 25(4):179–189. [find](https://scholar.google.com/scholar?q=Arafat+Awajan%2E+2007%2E+Arabic+text+preprocessing+for+the+natural+language+processing+applications%2E+Arab+Gulf+Journal+of+Scientific+Research%2C+25%284%29%3A179%E2%80%93189%2E)
14. Tanzil. 2023. Tanzil quran navigator. Accessed: 2023-10-09. [find](https://scholar.google.com/scholar?q=Tanzil%2E+2023%2E+Tanzil+quran+navigator%2E+Accessed%3A+2023%2D10%2D09%2E)
15. Kerstin Eksell. 1995. Complexity of linguistic change as reflected in arabic dialects. Studia Orientalia Electronica, 75:63–74. [find](https://scholar.google.com/scholar?q=Kerstin+Eksell%2E+1995%2E+Complexity+of+linguistic+change+as+reflected+in+arabic+dialects%2E+Studia+Orientalia+Electronica%2C+75%3A63%E2%80%9374%2E)
16. Lisa Torrey and Jude Shavlik. 2010. Transfer learning. In Handbook of research on machine learning applications and trends: algorithms, methods, and techniques, pages 242–264. IGI global. [find](https://scholar.google.com/scholar?q=Lisa+Torrey+and+Jude+Shavlik%2E+2010%2E+Transfer+learning%2E+In+Handbook+of+research+on+machine+learning+applications+and+trends%3A+algorithms%2C+methods%2C+and+techniques%2C+pages+242%E2%80%93264%2E+IGI+global%2E)
17. Christiane Fellbaum. 1998. Wordnet: An electronic lexical database. MIT Press google schola, 2:678– 686. [find](https://scholar.google.com/scholar?q=Christiane+Fellbaum%2E+1998%2E+Wordnet%3A+An+electronic+lexical+database%2E+MIT+Press+google+schola%2C+2%3A678%E2%80%93+686%2E)
18. Zineb Touati-Hamad, Mohamed Ridda Laouar, and Issam Bendib. 2021. Authentication of quran verses sequences using deep learning. In 2021 International Conference on Recent Advances in Mathematics and Informatics (ICRAMI), pages 1–4. IEEE. [find](https://scholar.google.com/scholar?q=Zineb+Touati%2DHamad%2C+Mohamed+Ridda+Laouar%2C+and+Issam+Bendib%2E+2021%2E+Authentication+of+quran+verses+sequences+using+deep+learning%2E+In+2021+International+Conference+on+Recent+Advances+in+Mathematics+and+I)
19. Anthony Gillioz, Jacky Casas, Elena Mugellini, and Omar Abou Khaled. 2020. Overview of the transformer-based models for nlp tasks. In 2020 15th Conference on computer science and information systems (FedCSIS), pages 179–183. IEEE. [find](https://scholar.google.com/scholar?q=Anthony+Gillioz%2C+Jacky+Casas%2C+Elena+Mugellini%2C+and+Omar+Abou+Khaled%2E+2020%2E+Overview+of+the+transformer%2Dbased+models+for+nlp+tasks%2E+In+2020+15th+Conference+on+computer+science+and+information+systems+%28)
20. Zineb Touati-Hamad, Mohamed Ridda Laouar, Issam Bendib, and Saqib Hakak. 2022. Arabic quran verses authentication using deep learning and word embeddings. The International Arab Journal of Information Technology, 19(4):681–688. [↗](https://doi.org/10.34028/iajit/19/4/13)
21. Naman Goyal, Jingfei Du, Myle Ott, Giri Anantharaman, and Alexis Conneau. 2021. [Larger-scale transformers for multilingual masked language modeling](https://md.research.elamur.ai/free/en/papers/goyal-2021-larger-scale-transformers-multilingual-masked/). arXiv preprint arXiv:2105.00572. [↗](https://arxiv.org/abs/2105.00572)
22. A Vaswani. 2017. [Attention is all you need](https://md.research.elamur.ai/free/en/papers/vaswani-2017-attention-all-need/). Advances in Neural Information Processing Systems.
23. Saqib Hakak, Amirrudin Kamsin, Omar Tayan, Mohd Yamani Idna Idris, and Gulshan Amin Gilkar. 2019. Approaches for preserving content integrity of sensitive online arabic content: A survey and research challenges. Information Processing & Management, 56(2):367–380. [find](https://scholar.google.com/scholar?q=Saqib+Hakak%2C+Amirrudin+Kamsin%2C+Omar+Tayan%2C+Mohd+Yamani+Idna+Idris%2C+and+Gulshan+Amin+Gilkar%2E+2019%2E+Approaches+for+preserving+content+integrity+of+sensitive+online+arabic+content%3A+A+survey+and+research+)
24. Zihan Wang, Stephen Mayhew, Dan Roth, et al. 2019. Cross-lingual ability of multilingual bert: An empirical study. arXiv preprint arXiv:1912.07840. [↗](https://arxiv.org/abs/1912.07840)

*Updated 2 November 2026*