Elamur Research⌕
☰
home / papers / 2022 / Detecting Users Prone to Spread Fake News on Arabic Twitter

Detecting Users Prone to Spread Fake News on Arabic Twitter

Paper · 2022 · Zien Sheikh Ali, Abdulaziz Al-Ali and Tamer Elsayed

Summary

The paper moves fake-news detection from single posts to the accounts that spread them. It asks whether an Arabic Twitter user is likely to circulate misinformation, whether on purpose or by accident. No Arabic resource existed for this, so the authors adapted the PAN 2020 user-labelling recipe. They started from professionally verified claims in ArCOV19-Rumors and AraFacts. They then found and manually checked tweets carrying those claims, gathered retweets, and searched each user's timeline for further claim matches with ElasticSearch. Users were labelled from how often they spread false claims. The final dataset has 1,546 users, 541 of them prone to spreading fake news. Each user is described by features from recent tweets and the profile: textual, statistical, emotional and account metadata. Several classifiers were compared, and logistic regression did best with an F1 of 0.73. Applied to the English PAN benchmark, the same approach beat the previous state of the art. This matters because rating sources can slow misinformation before each new post has to be checked.

Key points

  • Claim pool: 5,371 verified claims (299 True, 5,072 False), drawn from ArCOV19-Rumors (138 claims, 9,414 tweets) and AraFacts (6,222 claims from 5 fact-checking sites, filtered to True/False only).
  • Manual audit of AraFacts tweet–claim pairs gave 2,474 tweets expressing the claim, 125 Other and 7 negating, so about 95% of the fact-checkers' links were correct.
  • 35,698 retweets were collected, giving 4,176 unique candidate users. Up to 3,200 timeline tweets per user were searched against all claims with ElasticSearch.
  • The final ArPFN dataset has 1,546 users (541 prone). Logistic regression reached F1 = 0.73, and the approach also beat the state of the art on the English PAN 2020 data.
  • Profile features include rates normalised by account age, such as follower growth, tweet frequency and listed growth, plus counts of characters and digits in names and screen names. Code and features are public at gitlab.com/bigirqu/ArPFN.

Interesting ideas

  • I1. Stance-aware label inheritance: a tweet that denies a claim takes the opposite truth label instead of being thrown away, so stance controls how a label passes from a claim to the text that mentions it.
  • I2. A separate "questioning" category: tweets that only ask whether a claim is true are labelled Other, not counted as spreading it, which keeps asking apart from asserting.
  • I3. Audit inherited labels with a small manual pass: the authors re-checked links supplied by professionals and still found noise, including some negating tweets. Spot-checking upstream labels is cheap and reveals systematic errors.
  • I4. Densify sparse labels by retrieval: each user's whole timeline was searched for matches to every known claim, which found more evidence per user than the seed tweets alone.
  • I5. Normalise counts by exposure time: raw follower and tweet counts were turned into per-month rates based on account age, so old and new accounts can be compared fairly.

For Mova

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

  • I1 (frames): When frames link a sentence to a known proposition, the polarity of the speech act (asserting versus denying) should flip the inherited truth value before it reaches the world model.
  • I2 (frames): The speech-act layer should mark interrogative or doubting mentions so that they do not create belief or commitment entries for the speaker.
  • I3 (learn): Silver data from witness parsers or LLM teachers should get a small sampled audit with explicit "other" and "contradicting" buckets to estimate noise before training.
  • I4 (world): Mentions in a text could be retrieval-matched against stored propositions in the knowledge graph to attach extra known facts to entities.
  • I5 (world): Quantities tracked for entities could also be stored as rates over the relevant time span when comparison matters, not only as absolute totals.

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

 Fields, topics and concepts

📚 Shelves 6The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksSocial Sciences · 1,772 worksInformation Systems · 1,437 worksSociology and Political Science · 577 worksPhysics and Astronomy · 44 worksStatistical and Nonlinear Physics · 40 works

🧭 Topics 1Research subjects and languages the work is about.Semitic languages · 2,908 works

Cite

Cite as Zien Sheikh Ali, Abdulaziz Al-Ali and Tamer Elsayed. 2022. Detecting Users Prone to Spread Fake News on Arabic Twitter. Proceedinsg of the 5th Workshop on Open-Source Arabic Corpora and Processing Tools with Shared Tasks on Qur'an QA and Fine-Grained Hate Speech Detection.

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

@inproceedings{ali2022detecting,
  title = {{Detecting Users Prone to Spread Fake News on Arabic Twitter}},
  author = {Zien Sheikh Ali, Abdulaziz Al-Ali and Tamer Elsayed},
  booktitle = {Proceedinsg of the 5th Workshop on Open-Source Arabic Corpora and Processing Tools with Shared Tasks on Qur'an QA and Fine-Grained Hate Speech Detection},
  year = {2022},
  url = {https://aclanthology.org/2022.osact-1.2/}
}

References

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

  1. Abdul-Mageed, M., Elmadany, A., et al. (2021). Arbert & marbert: Deep bidirectional transformers for arabic. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers), pages 7088–7105. ↗
  2. An, J., Kwak, H., Lee, C. S., Jun, B., and Ahn, Y.-Y. (2021). Predicting anti-asian hateful users on twitter during covid-19. In Findings of the Association for Computational Linguistics: EMNLP 2021, pages 4655–4666. find
  3. Antoun, W., Baly, F., and Hajj, H. (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, pages 9–15. ↗
  4. Buda, J. and Bolonyai, F. (2020). An ensemble model using n-grams and statistical features to identify fake news spreaders on twitter. In CLEF. find
  5. Castillo, C., Mendoza, M., and Poblete, B. (2011). Information credibility on twitter. In Proceedings of the 20th international conference on World wide web, pages 675–684. find
  6. El Ballouli, R., El-Hajj, W., Ghandour, A., Elbassuoni, S., Hajj, H. M., and Shaban, K. B. (2017). Cat: Credibility analysis of arabic content on twitter. In WANLP@ EACL, pages 62–71.
  7. Fersini, E., Armanini, J., and D’Intorni, M. (2020). Profiling fake news spreaders: Stylometry, personality, emotions and embeddings. In CLEF (Working Notes). find
  8. Ghanem, B., Rosso, P., and Rangel, F. (2020). An emotional analysis of false information in social media and news articles. ACM Transactions on Internet Technology (TOIT), 20(2):1–18. ↗
  9. Haouari, F., Hasanain, M., Suwaileh, R., and Elsayed, T. (2021a). ArCOV-19: The First Arabic COVID-19 Twitter Dataset with Propagation Networks. In Proceedings of the Sixth Arabic Natural Language Processing Workshop, pages 82–91. ↗
  10. Haouari, F., Hasanain, M., Suwaileh, R., and Elsayed, T. (2021b). ArCOV19-Rumors: Arabic COVID-19 Twitter Dataset for Misinformation Detection. In Proceedings of the Sixth Arabic Natural Language Processing Workshop, pages 72–81. ↗
  11. Harrag, F. and Djahli, M. K. (2022). Arabic fake news detection: A fact checking based deep learning approach. Transactions on Asian and Low-Resource Language Information Processing, 21(4):1–34. ↗
  12. Hassan, S., Mubarak, H., Abdelali, A., and Darwish, K. (2021). Asad: Arabic social media analytics and understanding. In Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: System Demonstrations, pages 113–118.
  13. Islam, M. Z., Liu, J., Li, J., Liu, L., and Kang, W. (2019). A semantics aware random forest for text classification. In Proceedings of the 28th ACM International Conference on Information and Knowledge Management, pages 1061–1070. find
  14. Jones, M. O. (2019). The gulf information war— propaganda, fake news, and fake trends: The weaponization of twitter bots in the gulf crisis. International journal of communication, 13:27. find
  15. Jones, M. O. (2020). Disinformation superspreaders: the weaponisation of covid-19 fake news in the persian gulf and beyond. Global Discourse: An interdisciplinary journal of current affairs, 10(4):431– 437. ↗
  16. Lampos, V., Aletras, N., Preoţiuc-Pietro, D., and Cohn, T. (2014). Predicting and characterising user impact on twitter. In Proceedings of the 14th Conference of the European Chapter of the Association for Computational Linguistics, pages 405–413. find
  17. Mihaylov, T., Georgiev, G., and Nakov, P. (2015). Finding opinion manipulation trolls in news community forums. In Proceedings of the nineteenth conference on computational natural language learning, pages 310–314. find
  18. Moreno-Sandoval, L. G., Del Puertas, E. A. P., Quimbaya, A. P., and Alvarado-Valencia, J. A. (2020). Assembly of polarity, emotion and user statistics for detection of fake profiles. In CLEF (Working Notes). find
  19. Nakov, P., Da San Martino, G., Elsayed, T., Barrón-Cedeño, A., Mı́guez, R., Shaar, S., Alam, F., Haouari, F., Hasanain, M., Mansour, W., et al. (2021). Overview of the clef–2021 checkthat! lab on detecting check-worthy claims, previously fact-checked claims, and fake news. In International Conference of the Cross-Language Evaluation Forum for European Languages, pages 264– 291. Springer. find
  20. Rangel, F., Giachanou, A., Ghanem, B., and Rosso, P. (2020). Overview of the 8th author profiling task at pan 2020: Profiling fake news spreaders on twitter. In CLEF. find
  21. Rath, B., Salecha, A., and Srivastava, J. (2020). Detecting fake news spreaders in social networks using inductive representation learning. In 2020 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM), pages 182–189. IEEE. find
  22. Reimers, N. and Gurevych, I. (2019). Sentence-bert: Sentence embeddings using siamese bert-networks. 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 3982– 3992.
  23. Shao, C., Ciampaglia, G. L., Varol, O., Yang, K.-C., Flammini, A., and Menczer, F. (2018). The spread of low-credibility content by social bots. Nature communications, 9(1):1–9. ↗
  24. Sheikh Ali, Z., Mansour, W., Elsayed, T., and Al-Ali, A. (2021). AraFacts: The First Large Arabic Dataset of Naturally-Occurring Professionally-Verified Claims. In Proceedings of the Sixth Arabic Natural Language Processing Workshop. Association for Computational Linguistics. find
  25. Shu, K., Mahudeswaran, D., Wang, S., Lee, D., and Liu, H. (2018). FakeNewsNet: A data repository with news content, social context and spatialtemporal information for studying fake news on social media. arXiv preprint arXiv:1809.01286. ↗
  26. Shu, K., Zhou, X., Wang, S., Zafarani, R., and Liu, H. (2019). The role of user profiles for fake news detection. In Proceedings of the 2019 IEEE/ACM international conference on advances in social networks analysis and mining, pages 436–439. find
  27. Shu, K., Bhattacharjee, A., Alatawi, F., Nazer, T. H., Ding, K., Karami, M., and Liu, H. (2020). Combating disinformation in a social media age. Wiley Interdisciplinary Reviews: Data Mining and Knowledge Discovery, 10(6):e1385. ↗
  28. Shwartz-Ziv, R. and Armon, A. (2022). Tabular data: Deep learning is not all you need. Information Fusion, 81:84–90. ↗
  29. Vosoughi, S., Roy, D., and Aral, S. (2018). The spread of true and false news online. Science, 359(6380):1146–1151. ↗
  30. Yang, K.-C., Varol, O., Hui, P.-M., and Menczer, F. (2020). Scalable and generalizable social bot detection through data selection. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 34, pages 1096–1103. find
  31. Zhang, X., Cao, J., Li, X., Sheng, Q., Zhong, L., and Shu, K. (2021). Mining dual emotion for fake news detection. In Proceedings of the Web Conference 2021, pages 3465–3476. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/ali-2022-detecting-users-prone-spread-fake/