---
title: "AraFacts — paper by Ali et al. (2021) — summary and ideas"
description: "Automated fact-checking for Arabic has been held back by the lack of suitable data. 3 references, cited by 16 papers here."
authors: ["Zien Sheikh Ali", "Watheq Mansour", "Tamer Elsayed", "Abdulaziz Al-Ali"]
year: 2021
venue: "Proceedings of the Sixth Arabic Natural Language Processing Workshop"
source: "https://aclanthology.org/2021.wanlp-1.26/"
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/ali-2021-arafacts-first-large-arabic-dataset/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/ali-2021-arafacts-first-large-arabic-dataset/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2021](https://md.research.elamur.ai/free/en/papers/2021/page/6/) / AraFacts: The First Large Arabic Dataset of Naturally Occurring Claims

# AraFacts: The First Large Arabic Dataset of Naturally Occurring Claims

**Authors:** [Zien Sheikh Ali](https://md.research.elamur.ai/free/en/authors/zien-sheikh-ali/), [Watheq Mansour](https://md.research.elamur.ai/free/en/authors/watheq-mansour/), [Tamer Elsayed](https://md.research.elamur.ai/free/en/authors/tamer-elsayed/) and [Abdulaziz Al-Ali](https://md.research.elamur.ai/free/en/authors/abdulaziz-al-ali/) · **Year:** 2021 · **Venue:** [Proceedings of the Sixth Arabic Natural Language Processing Workshop](https://md.research.elamur.ai/free/en/venues/wanlp/) ([WANLP 2021](https://md.research.elamur.ai/free/en/venues/wanlp/2021/)) · **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/2021.wanlp-1.26/) · [PDF](https://aclanthology.org/2021.wanlp-1.26.pdf) · [OpenAlex](https://openalex.org/W3156584161)

**Citations:** 11 citations (OpenAlex, 2026-10-14) · [16](https://md.research.elamur.ai/free/en/papers/ali-2021-arafacts-first-large-arabic-dataset/cited-by/) in our library, of the 149,439 papers whose bibliographies we read

## Summary
Automated fact-checking for Arabic has been held back by the lack of suitable data. Several Arabic fact-checking outlets publish verdicts written by journalists, but their results are scattered across sites that each use their own rating scale. The authors built AraFacts by crawling five such sites and collecting 6,222 claims dating from 2016 onward. Each claim comes with its verdict, a description, the full fact-checking article, links to posts that spread it, links to evidence, a topic, a date and a type (text, image or video). Because every site rates claims differently, they mapped 27 original verdict labels onto four shared ones and 35 site-specific topics onto eight shared categories. Both the original and the normalized values are stored side by side. Their analysis shows that false claims dominate every topic, that most claims are about politics or news, and that Facebook and Twitter are the main places these claims start. The dataset is public and supports claim verification, detection of already-checked claims, and evidence retrieval.

## Key points
- Claims per source: Misbar 2,952, Fatabyyano 1,503, FactuelAFP (AFP's Arabic service) 973, Verify-Sy 707, Maharat-news 87. Sites were chosen because they are IFCN-certified or widely followed in the region.
- Normalized verdicts are False, Partly-False, Sarcasm and True. The eight shared topic categories cover Health, Religion, Politics, Arts & Culture, News, Social and General Sciences, plus a catch-all Other. Two authors did the mapping independently and then settled their disagreements.
- Each record has a 13-field schema, including Source-label next to Normalized-label, Claim URLs, Evidence URLs and Claim type. The text of the fact-checking article is also included.
- Facebook accounts for 54.5% of claim-origin URLs and Twitter for 25.0%. Most claims date from 2020. Misbar and Fatabyyano, the two largest sources, contain more image and video claims than text claims.
- Verify-Sy gives no explicit verdicts. The authors read its editorial policy, checked a sample by hand, and concluded that every claim on the site is false.

## Interesting ideas
- Store the original label next to the unified one instead of replacing it, so information is not lost and the fine-grained labels can serve as an extra multi-task training signal.
- Have two people build a mapping from many source schemes to one target scheme independently, then compare and resolve the differences, which shows where the mapping is ambiguous.
- When a source never states a label, infer it from the source's declared policy and then confirm by checking a sample by hand, rather than trusting the policy alone.
- Before verifying a new claim, check whether the same claim has already been checked, since much misinformation is simply repeated.

## For Mova

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

- **I1** (tree): When converting treebanks between UD dialects, keep each source tag and relation alongside the converted one, so every conversion decision can be traced and undone.
- **I2** (world): Mapping WordNet, FrameNet and Wikidata categories into one ontology could follow the same pattern, with two independent mappings (human or LLM) and only the disagreements reviewed.
- **I3** (learn): Silver data taken from corpora with known conventions can be given default labels from those conventions, as long as a manual sample check is done before training.
- **I4** (world): Before adding an assertion to a text's world, look for an existing matching state or event so repeated statements reinforce one entry instead of creating duplicates.

*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 (6): [🖥️ 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 · [🤝 Sociology and Political Science](https://md.research.elamur.ai/free/en/shelves/social-sciences/sociology-and-political-science/) · 577 works

🧭 Topics (1): [Semitic languages](https://md.research.elamur.ai/free/en/topics/semitic-languages/) · 2,908 works

## Cite

Cite as: Zien Sheikh Ali, Watheq Mansour, Tamer Elsayed and Abdulaziz Al-Ali. 2021. AraFacts: The First Large Arabic Dataset of Naturally Occurring Claims. Proceedings of the Sixth Arabic Natural Language Processing Workshop.

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

```bibtex
@inproceedings{ali2021arafacts,
  title = {{AraFacts: The First Large Arabic Dataset of Naturally Occurring Claims}},
  author = {Zien Sheikh Ali, Watheq Mansour, Tamer Elsayed and Abdulaziz Al-Ali},
  booktitle = {Proceedings of the Sixth Arabic Natural Language Processing Workshop},
  year = {2021},
  url = {https://aclanthology.org/2021.wanlp-1.26/}
}
```

- [← previous Arabic Code-Switching Speech Recognition using Monolingual Data](https://md.research.elamur.ai/free/en/papers/ali-2021-arabic-code-switching-speech-recognition/)
- [next → ChEMU-Ref: A Corpus for Modeling Anaphora Resolution in the Chemical Domain](https://md.research.elamur.ai/free/en/papers/fang-2021-chemu-ref-corpus-modeling-anaphora/)
- [↑ back to papers of 2021](https://md.research.elamur.ai/free/en/papers/2021/page/6/)

## References

3 references, 2 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Isabelle Augenstein, Christina Lioma, Dongsheng Wang, Lucas Chaves Lima, Casper Hansen, Christian Hansen, and Jakob Grue Simonsen. 2019. [Multifc: A real-world multi-domain dataset for evidence-based fact checking of claims](https://md.research.elamur.ai/free/en/papers/augenstein-2019-multifc-real-world-multi-domain/). 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 4677–4691. [↗](https://doi.org/10.48550/arxiv.1909.03242)
2. Ramy Baly, Mitra Mohtarami, and James Glass. 2018. [Integrating stance detection and fact checking in a unified corpus](https://md.research.elamur.ai/free/en/papers/baly-2018-integrating-stance-detection-fact-checking/). In Proceedings of NAACL-HLT, pages 21–27.
3. Image-based Fact-checking Alberto Barrón-Cedeno, Tamer Elsayed, Preslav Nakov, Giovanni Da San Martino, Maram Hasanain, Reem Suwaileh, Fatima Haouari, Nikolay Babulkov, Bayan Hamdan, Alex Nikolov, et al. 2020. Overview of checkthat! 2020: Automatic identification and verification of claims in social media. In International Conference of the Cross-Language Evaluation Forum for European Languages, pages 215– 236. Springer. [find](https://scholar.google.com/scholar?q=Image%2Dbased+Fact%2Dchecking+Alberto+Barro%CC%81n%2DCedeno%2C+Tamer+Elsayed%2C+Preslav+Nakov%2C+Giovanni+Da+San+Martino%2C+Maram+Hasanain%2C+Reem+Suwaileh%2C+Fatima+Haouari%2C+Nikolay+Babulkov%2C+Bayan+Hamdan%2C+Alex+Nikolov%2C+et)

*Updated 1 November 2026*