---
title: "COVID-19 and Arabic Twitter: How can Arab World Governments"
description: "The paper looks at how Arabic-language Twitter could help health authorities in Arab… 84% accuracy. 25 references, cited by 5 papers here."
authors: ["Lama Alsudias", "Paul Rayson"]
year: 2020
venue: "Proceedings of the 1st Workshop on NLP for COVID-19 at ACL 2020"
source: "https://aclanthology.org/2020.nlpcovid19-acl.16/"
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/alsudias-2020-covid-19-arabic-twitter-how/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/alsudias-2020-covid-19-arabic-twitter-how/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2020](https://md.research.elamur.ai/free/en/papers/2020/page/6/) / COVID-19 and Arabic Twitter: How can Arab World Governments and Public Health Organizations Learn from Social Media?

# COVID-19 and Arabic Twitter: How can Arab World Governments and Public Health Organizations Learn from Social Media?

**Authors:** [Lama Alsudias](https://md.research.elamur.ai/free/en/authors/lama-alsudias/) and [Paul Rayson](https://md.research.elamur.ai/free/en/authors/paul-rayson/) · **Year:** 2020 · **Venue:** [Proceedings of the 1st Workshop on NLP for COVID-19 at ACL 2020](https://md.research.elamur.ai/free/en/venues/nlpcovid19/) ([Workshop on NLP for COVID-19 at ACL 2020](https://md.research.elamur.ai/free/en/venues/nlpcovid19/2020/)) · **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/2020.nlpcovid19-acl.16/) · [PDF](https://aclanthology.org/2020.nlpcovid19-acl.16.pdf) · [OpenAlex](https://openalex.org/W3082108391)

**Citations:** 44 citations (OpenAlex, 2026-10-14) · [5](https://md.research.elamur.ai/free/en/papers/alsudias-2020-covid-19-arabic-twitter-how/cited-by/) in our library, of the 149,439 papers whose bibliographies we read

## Summary
The paper looks at how Arabic-language Twitter could help health authorities in Arab countries during COVID-19. Earlier pandemic social-media work focused almost entirely on English. The authors first extended their Arabic Infectious Disease Ontology with COVID-19 vocabulary, including slang. They then gathered Arabic tweets from December 2019 to April 2020 using three keywords, one of them a common misspelling of "Corona". After cleaning, about a million unique tweets remained.

They ran three analyses:
- **Topic discovery:** k-means clustering over n-grams.
- **Rumour detection:** classifiers trained on a hand-labelled sample of 2,000 tweets.
- **Source-type prediction:** a classifier from their earlier work, applied to the tweets flagged as false.

Simple word-count features with linear classifiers reached about 84% accuracy on rumour detection. Word embeddings did far worse. Most false tweets were predicted to come from academic-sounding or health-professional-sounding writers.

This matters because the gap between a rumour appearing and an official rebuttal can be months long. Fast automatic flagging in Arabic, including dialects, could shorten that delay.

## Key points
- **Corpus:** 6,578,982 tweets were collected weekly via the Twitter API and reduced to 1,048,575 unique tweets. Cleaning removed retweets and spam, stripped URLs, hashtags and emojis, normalized the text and dropped stopwords.
- **Clustering:** k-means with k=5 on unigrams, bigrams and trigrams (scikit-learn 0.20.2) produced five themes:
  - case statistics,
  - prayers,
  - locations of outbreaks,
  - prevention advice,
  - advertising.
- **Gold labels:** the 2,000-tweet sample split into 316 false, 895 true and 789 unrelated. False tweets are about 26% of the related tweets.
- **Classifier results:** the best accuracy was 84.03%, reached both by Logistic Regression with count vectors and by SVC with TF-IDF. Word2Vec and FastText features scored only 50–60% accuracy, with F1 around 40%.
- **Source types:** of the 316 false tweets, 30% were predicted as health professional and 28% as academic, but only 12% as public. The dataset, tweet IDs and labels are released for research use.

## Interesting ideas
- The ontology pairs formal medical terms with their everyday and slang counterparts, and that mapping also drives data collection, so informal posts are not missed.
- Deliberately searching for a frequent misspelling alongside the correct spelling captures a large slice of real usage that strict matching would lose.
- Rumour annotation is done top-down: known false claims are fixed in advance, then matching posts are sampled, giving a controlled and checkable label set.
- The finding that false claims often wear an expert-sounding style shows that how authoritative a text sounds is separate from whether its content is true.
- Plain count and TF-IDF baselines beat embeddings by roughly 20 points on a small domain-specific set, a reminder that cheap baselines can win in low-data settings.

## For Mova

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

- **I1** (world): Our WordNet/Wikidata-backed ontology could store register variants (colloquial, slang, technical) as aliases of one concept, so entities in informal text resolve to the same node as formal mentions.
- **I2** (tokens): The en lemmatizer could keep a curated table of high-frequency misspellings mapped to canonical lemmas, with a flag kept for explainability.
- **I3** (learn): Evaluation sets could be assembled around a predefined inventory of target phenomena and then filled with sampled sentences, which makes coverage gaps and negative controls explicit.
- **I4** (frames): prag could model source framing such as "a Chinese expert confirms" or "in scientific reading" as an evidential or authority speech act, kept separate from the truth status of the embedded claim in the world model.
- **I5** (learn): Every new learned component should be reported against a trivial frequency baseline, so gains from heavier models are shown rather than assumed.

*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 (3): [🤝 Social Sciences](https://md.research.elamur.ai/free/en/shelves/social-sciences/) · 1,954 works · [🤝 Sociology and Political Science](https://md.research.elamur.ai/free/en/shelves/social-sciences/sociology-and-political-science/) · 626 works · [🤝 Communication](https://md.research.elamur.ai/free/en/shelves/social-sciences/communication/) · 104 works

🧭 Topics (1): [Semitic languages](https://md.research.elamur.ai/free/en/topics/semitic-languages/) · 3,066 works

## Cite

Cite as: Lama Alsudias and Paul Rayson. 2020. COVID-19 and Arabic Twitter: How can Arab World Governments and Public Health Organizations Learn from Social Media?. Proceedings of the 1st Workshop on NLP for COVID-19 at ACL 2020.

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

```bibtex
@inproceedings{alsudias2020covid,
  title = {{COVID-19 and Arabic Twitter: How can Arab World Governments and Public Health Organizations Learn from Social Media?}},
  author = {Lama Alsudias and Paul Rayson},
  booktitle = {Proceedings of the 1st Workshop on NLP for COVID-19 at ACL 2020},
  year = {2020},
  url = {https://aclanthology.org/2020.nlpcovid19-acl.16/}
}
```

- [← previous Event Coreference Resolution with their Paraphrases and Argument-aware Embeddings](https://md.research.elamur.ai/free/en/papers/zeng-2020-event-coreference-resolution-their-paraphrases/)
- [next → Verum focus is verum, not focus: Cross-linguistic evidence](https://md.research.elamur.ai/free/en/papers/gutzmann-2020-verum-focus-is-verum-not/)
- [↑ back to papers of 2020](https://md.research.elamur.ai/free/en/papers/2020/page/6/)

## References

25 references, 4 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Wasim Ahmed, Peter A Bath, Laura Sbaffi, and Gianluca Demartini. 2019. Novel insights into views towards H1N1 during the 2009 Pandemic: a thematic analysis of Twitter data. Health Information & Libraries Journal, 36(1):60–72. [find](https://scholar.google.com/scholar?q=Wasim+Ahmed%2C+Peter+A+Bath%2C+Laura+Sbaffi%2C+and+Gianluca+Demartini%2E+2019%2E+Novel+insights+into+views+towards+H1N1+during+the+2009+Pandemic%3A+a+thematic+analysis+of+Twitter+data%2E+Health+Information+%26+Librar)
2. Wasim Ahmed, Gianluca Demaerini, and Peter A Bath. 2017a. Topics discussed on twitter at the beginning of the 2014 Ebola epidemic in United States. iConference 2017 Proceedings. [find](https://scholar.google.com/scholar?q=Wasim+Ahmed%2C+Gianluca+Demaerini%2C+and+Peter+A+Bath%2E+2017a%2E+Topics+discussed+on+twitter+at+the+beginning+of+the+2014+Ebola+epidemic+in+United+States%2E+iConference+2017+Proceedings%2E)
3. Wasim Ahmed, G. Demartini, and P. Bath. 2017b. Topics Discussed on Twitter at the Beginning of the 2014 Ebola Epidemic in United States. [find](https://scholar.google.com/scholar?q=Wasim+Ahmed%2C+G%2E+Demartini%2C+and+P%2E+Bath%2E+2017b%2E+Topics+Discussed+on+Twitter+at+the+Beginning+of+the+2014+Ebola+Epidemic+in+United+States%2E)
4. Sarah Alqurashi, Ahmad Alhindi, and Eisa Alanazi. 2020. [Large arabic twitter dataset on covid-19](https://md.research.elamur.ai/free/en/papers/alqurashi-2020-large-arabic-twitter-dataset-covid/). arXiv preprint arXiv:2004.04315. 11 https://doi.org/10.17635/lancaster/ researchdata/375 [↗](https://arxiv.org/abs/2004.04315)
5. Mohamed Taher Alrefaie. 2017. arabic stop words. [https://github.com/mohataher/](https://github.com/mohataher/) arabic-stop-words . [↗](https://github.com/mohataher/)
6. Lama Alsudias and Paul Rayson. 2019. [Classifying information sources in Arabic twitter to support online monitoring of infectious diseases](https://md.research.elamur.ai/free/en/papers/alsudias-2019-classifying-information-sources-arabic-twitter/). In Proceedings of the 3rd Workshop on Arabic Corpus Linguistics, pages 22–30, Cardiff, United Kingdom. Association for Computational Linguistics.
7. Lama Alsudias and Paul Rayson. 2020. [Developing an Arabic Infectious Disease Ontology to Include Non-Standard Terminology](https://md.research.elamur.ai/free/en/papers/alsudias-2020-developing-arabic-infectious-disease-ontology/). In Proceedings of The 12th Language Resources and Evaluation Conference, pages 4844–4852, Marseille, France. European Language Resources Association.
8. Piotr Bojanowski, Edouard Grave, Armand Joulin, and Tomas Mikolov. 2017. [Enriching word vectors with subword information](https://md.research.elamur.ai/free/en/papers/bojanowski-2017-enriching-word-vectors-subword-information/). Transactions of the Association for Computational Linguistics, 5:135–146. [↗](https://doi.org/10.1162/tacl_a_00051)
9. Emily Chen, Kristina Lerman, and Emilio Ferrara. 2020. Covid-19: The first public coronavirus twitter dataset. arXiv preprint arXiv:2003.07372. [↗](https://arxiv.org/abs/2003.07372)
10. I.C.-H Fung, King-wa Fu, C.-H Chan, Benedict Chan, Chi-Ngai Cheung, Thomas Abraham, and Zion Tse. 2016a. Social Media’s Initial Reaction to Information and Misinformation on Ebola, August 2014: Facts and Rumors. Public Health Reports, 131:461– 473. [find](https://scholar.google.com/scholar?q=I%2EC%2E%2DH+Fung%2C+King%2Dwa+Fu%2C+C%2E%2DH+Chan%2C+Benedict+Chan%2C+Chi%2DNgai+Cheung%2C+Thomas+Abraham%2C+and+Zion+Tse%2E+2016a%2E+Social+Media%E2%80%99s+Initial+Reaction+to+Information+and+Misinformation+on+Ebola%2C+August+2014%3A+Facts+)
11. Isaac Chun-Hai Fung, Carmen Hope Duke, Kathryn Cameron Finch, Kassandra Renee Snook, Pei-Ling Tseng, Ana Cristina Hernandez, Manoj Gambhir, King-Wa Fu, and Zion Tsz Ho Tse. 2016b. Ebola virus disease and social media: A systematic review. American journal of infection control, 44(12):1660—1671. [find](https://scholar.google.com/scholar?q=Isaac+Chun%2DHai+Fung%2C+Carmen+Hope+Duke%2C+Kathryn+Cameron+Finch%2C+Kassandra+Renee+Snook%2C+Pei%2DLing+Tseng%2C+Ana+Cristina+Hernandez%2C+Manoj+Gambhir%2C+King%2DWa+Fu%2C+and+Zion+Tsz+Ho+Tse%2E+2016b%2E+Ebola+virus+disease+)
12. Amira Ghenai and Yelena Mejova. 2017. Catching Zika fever: Application of crowdsourcing and machine learning for tracking health misinformation on Twitter. arXiv preprint arXiv:1707.03778. [↗](https://arxiv.org/abs/1707.03778)
13. Fatima Haouari, Maram Hasanain, Reem Suwaileh, and Tamer Elsayed. 2020. The First Arabic COVID-19 Twitter Dataset with Propagation Networks. arXiv preprint arXiv:2004.05861. [↗](https://arxiv.org/abs/2004.05861)
14. Aditya Joshi, Sarvnaz Karimi, Ross Sparks, Cécile Paris, and C Raina Macintyre. 2019. Survey of Text-based Epidemic Intelligence: A Computational Linguistics Perspective. ACM Computing Surveys (CSUR), 52(6):1–19. [find](https://scholar.google.com/scholar?q=Aditya+Joshi%2C+Sarvnaz+Karimi%2C+Ross+Sparks%2C+Ce%CC%81cile+Paris%2C+and+C+Raina+Macintyre%2E+2019%2E+Survey+of+Text%2Dbased+Epidemic+Intelligence%3A+A+Computational+Linguistics+Perspective%2E+ACM+Computing+Surveys+%28CSUR%29)
15. Janani Kalyanam, Sumithra Velupillai, Son Doan, Mike Conway, and Gert R. G. Lanckriet. 2015. Facts and Fabrications about Ebola: A Twitter Based Study. ArXiv, abs/1508.02079. [↗](https://arxiv.org/abs/1508.02079)
16. Christian E Lopez, Malolan Vasu, and Caleb Gallemore. 2020. Understanding the perception of COVID-19 policies by mining a multilanguage Twitter dataset. arXiv preprint arXiv:2003.10359. [↗](https://arxiv.org/abs/2003.10359)
17. Edward Ma. 2018. 3 basic approaches in Bag of Words which are better than Word Embeddings. [find](https://scholar.google.com/scholar?q=Edward+Ma%2E+2018%2E+3+basic+approaches+in+Bag+of+Words+which+are+better+than+Word+Embeddings%2E)
18. Rabeaa Mouty and Achraf Gazdar. 2018. Survey on Steps of Truth Detection on Arabic Tweets. In 2018 21st Saudi Computer Society National Computer Conference (NCC), pages 1–6. IEEE. [find](https://scholar.google.com/scholar?q=Rabeaa+Mouty+and+Achraf+Gazdar%2E+2018%2E+Survey+on+Steps+of+Truth+Detection+on+Arabic+Tweets%2E+In+2018+21st+Saudi+Computer+Society+National+Computer+Conference+%28NCC%29%2C+pages+1%E2%80%936%2E+IEEE%2E)
19. Michelle Odlum and Sunmoo Yoon. 2015. What can we learn about the Ebola outbreak from tweets? American journal of infection control, 43(6):563– 571. [find](https://scholar.google.com/scholar?q=Michelle+Odlum+and+Sunmoo+Yoon%2E+2015%2E+What+can+we+learn+about+the+Ebola+outbreak+from+tweets%3F+American+journal+of+infection+control%2C+43%286%29%3A563%E2%80%93+571%2E)
20. Yeimer Ortiz-Martı́nez and Luisa F Jiménez-Arcia. 2017. Yellow fever outbreaks and Twitter: Rumors and misinformation. American journal of infection control, 45(7):816–817. Fabian Pedregosa, Gaël. [find](https://scholar.google.com/scholar?q=Yeimer+Ortiz%2DMart%C4%B1%CC%81nez+and+Luisa+F+Jime%CC%81nez%2DArcia%2E+2017%2E+Yellow+fever+outbreaks+and+Twitter%3A+Rumors+and+misinformation%2E+American+journal+of+infection+control%2C+45%287%29%3A816%E2%80%93817%2E+Fabian+Pedregosa%2C+Gae%CC%88l%2E)
21. Varoquaux, Alexandre Gramfort, Vincent Michel, Bertrand Thirion, Olivier Grisel, Mathieu Blondel, Peter Prettenhofer, Ron Weiss, Vincent Dubourg, Jake Vanderplas, Alexandre Passos, David Cournapeau, Matthieu Brucher, Matthieu Perrot, and Édouard Duchesnay. 2011. Scikit-learn: Machine Learning in Python. Journal of Machine Learning Research, 12:2825–2830. [find](https://scholar.google.com/scholar?q=Varoquaux%2C+Alexandre+Gramfort%2C+Vincent+Michel%2C+Bertrand+Thirion%2C+Olivier+Grisel%2C+Mathieu+Blondel%2C+Peter+Prettenhofer%2C+Ron+Weiss%2C+Vincent+Dubourg%2C+Jake+Vanderplas%2C+Alexandre+Passos%2C+David+Cournapeau%2C+M)
22. Umair Qazi, Muhammad Imran, and Ferda Ofli. 2020. GeoCoV19: A Dataset of Hundreds of Millions of Multilingual COVID-19 Tweets with Location Information. arXiv preprint arXiv:2005.11177. [↗](https://arxiv.org/abs/2005.11177)
23. Tatiana Vorovchenko, Proochista Ariana, Francois van Loggerenberg, and Amirian Pouria. 2017. Ebola and Twitter. What Insights Can Global Health Draw from Social Media?, pages 85–98. [find](https://scholar.google.com/scholar?q=Tatiana+Vorovchenko%2C+Proochista+Ariana%2C+Francois+van+Loggerenberg%2C+and+Amirian+Pouria%2E+2017%2E+Ebola+and+Twitter%2E+What+Insights+Can+Global+Health+Draw+from+Social+Media%3F%2C+pages+85%E2%80%9398%2E)
24. Arkaitz Zubiaga, Ahmet Aker, Kalina Bontcheva, Maria Liakata, and Rob Procter. 2018a. Detection and Resolution of Rumours in Social Media: A Survey. ACM Comput. Surv., 51(2). [find](https://scholar.google.com/scholar?q=Arkaitz+Zubiaga%2C+Ahmet+Aker%2C+Kalina+Bontcheva%2C+Maria+Liakata%2C+and+Rob+Procter%2E+2018a%2E+Detection+and+Resolution+of+Rumours+in+Social+Media%3A+A+Survey%2E+ACM+Comput%2E+Surv%2E%2C+51%282%29%2E)
25. Arkaitz Zubiaga, Ahmet Aker, Kalina Bontcheva, Maria Liakata, and Rob Procter. 2018b. Detection and resolution of rumours in social media: A survey. ACM Computing Surveys (CSUR), 51(2):1–36. [find](https://scholar.google.com/scholar?q=Arkaitz+Zubiaga%2C+Ahmet+Aker%2C+Kalina+Bontcheva%2C+Maria+Liakata%2C+and+Rob+Procter%2E+2018b%2E+Detection+and+resolution+of+rumours+in+social+media%3A+A+survey%2E+ACM+Computing+Surveys+%28CSUR%29%2C+51%282%29%3A1%E2%80%9336%2E)

*Updated 2 November 2026*