Elamur Research⌕
☰
home / papers / 2026 / Grounding Information Disorder in NLP: A Theoretical and Operational Framework

Grounding Information Disorder in NLP: A Theoretical and Operational Framework

Paper · 2026 · Wajdi Zaghouani

Summary

This position paper argues that treating fake news as a simple true/false problem cannot capture the variety of misleading content. Satire mistaken for fact, real images with deceptive captions, decontextualized statistics, and deliberate fabrications each call for different responses. Zaghouani proposes an NLP framework that assesses content along three linked axes: epistemic status (what verification can establish), intentionality (signs of manipulation or coordinated deception), and contextual harm (likely consequences given topic, timing, audience and reach).

The framework is a four-stage pipeline:

  1. Split content into atomic claims.
  2. Assess each claim in three parallel streams.
  3. Merge the results with formal aggregation operators that include conflict-resolution rules.
  4. Route outputs to actions using standardized risk profiles.

The paper also outlines a new evaluation protocol. Its central idea is to judge whether the system actually helps human decision-makers, not just how accurate the classifier is. No experiments are reported, and the author names empirical validation as essential future work. The contribution is a roadmap that draws together harm taxonomies, uncertainty quantification and fact-checking research to make misinformation assessment interpretable and aligned with policy.

Key points

  • Six stated contributions:
    • annotation schemas with rules for ambiguous cases;
    • aggregation operators with monotonicity and escalation guarantees;
    • operationalization of the FABLE harm taxonomy;
    • conformal prediction for abstention;
    • human uplift assessment;
    • risk profile templates.
  • FABLE (Sehat et al., 2024) is the primary harm taxonomy. Its five dimensions are Fragmentation, Actionability, Believability, Likelihood of spread, and Exploitativeness.
  • The intentionality stream relies on proxies (source history, linguistic manipulation markers, coordination signals). This choice follows evidence that intent cannot be reliably inferred from text alone.
  • The proposed evaluation covers conformal abstention, calibration analysis, modeling of annotator disagreement, harm-weighted metrics, and human uplift studies.
  • The paper builds on existing resources such as FEVER (185K claims), LIAR (six-level scale), CheckThat!, MuMiN (41 languages), X-FACT, MM-COVID, VERITE and NewsCLIPpings. It stresses that harm judgments vary across cultures.

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

🎓Northwestern University in QatarDoha · Qatar39 works · 148 citations

 Fields, topics and concepts

📚 Shelves 5The field of science this work belongs to (OpenAlex fields).Computer Science · 18,614 worksArtificial Intelligence · 17,965 worksSocial Sciences · 1,954 worksInformation Systems · 1,730 worksSociology and Political Science · 626 works

Cite

Cite as Wajdi Zaghouani. 2026. Grounding Information Disorder in NLP: A Theoretical and Operational Framework. Proceedings of the 1st Workshop on Information Disorder (InDor) @ LREC 2026.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.63317/23yq8ix5asa6

@inproceedings{zaghouani2026grounding,
  title = {{Grounding Information Disorder in NLP: A Theoretical and Operational Framework}},
  author = {Wajdi Zaghouani},
  booktitle = {Proceedings of the 1st Workshop on Information Disorder (InDor) @ LREC 2026},
  year = {2026},
  doi = {10.63317/23yq8ix5asa6},
  url = {https://doi.org/10.63317/23yq8ix5asa6}
}

References

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

  1. shared task: Persuasion techniques and disinformation detection in Arabic text. arXiv preprint arXiv:2311.03179. — [arXiv:2311.03179](https://arxiv.org/abs/2311.03179) find
  2. Panda, S. and Levitan, S. I. (2021). Detecting multilingual COVID-19 misinformation via contextualized embeddings. In Proceedings of NLP4IF Workshop, pp. 125–129. find
  3. Hasanain, M., Suwaileh, R., Weering, S., Li, C., Caselli, T., Zaghouani, W., Barrón-Cedeño, A., Nakov, P., and Alam, F. (2024). Overview of the CLEF-2024 CheckThat! lab task 1 on check-worthiness estimation of multigenre content. In Working Notes of CLEF 2024, pp. 276–286. find
  4. Papadopoulos, S.-I., Koutlis, C., Papadopoulos, S., and Petrantonakis, P. C. (2024). VERITE: A robust benchmark for multimodal misinformation detection accounting for unimodal bias. Int. J. Multimedia Information Retrieval, 13:4. find
  5. Horne, B. D. and Adali, S. (2017). This just in: Fake news packs a lot in title, uses simpler, repetitive content. In Proceedings of ICWSM, 11(1):759– 766. find
  6. Piot, P., Martín-Rodilla, P., and Parapar, J. (2025). Personalisation or prejudice? Addressing geographic bias in hate speech detection using debias tuning in large language models. arXiv preprint arXiv:2505.02252. — [arXiv:2505.02252](https://arxiv.org/abs/2505.02252) find
  7. Devic, S., Srinivasan, T., Thomason, J., Neiswanger, W., and Sharan, V. (2025). From calibration to collaboration: LLM uncertainty quantification should be more human-centered. arXiv preprint arXiv:2506.07461. — [arXiv:2506.07461](https://arxiv.org/abs/2506.07461) find
  8. Plaza-del-Arco, F. M., Cercas Curry, A., Paoli, S., Cercas Curry, A., and Hovy, D. (2024). Divine LLaMAs: Bias, stereotypes, stigmatization, and emotion representation of religion in large language models. In Findings of EMNLP 2024, pp. 4346–4366. find
  9. Li, Y., Jiang, B., Shu, K., and Liu, H. (2020). MM-COVID: A multilingual and multimodal data repository for combating COVID-19 disinformation. arXiv preprint arXiv:2011.04088. — [arXiv:2011.04088](https://arxiv.org/abs/2011.04088) find
  10. Prabhudesai, S., et al. (2023). Understanding uncertainty: How lay decision makers perceive uncertainty in human AI decision making. In Proceedings of IUI, pp. 379–396. find
  11. Luo, G., Darrell, T., and Rohrbach, A. (2021). NewsCLIPpings: Automatic generation of out of context multimodal media. In Proceedings of EMNLP, pp. 6801–6817. find
  12. Quach, V., Fisch, A., Schuster, T., et al. (2023). Conformal language modeling. In Proceedings of ICLR. find
  13. MacAvaney, S., et al. (2019). Hate speech detection: Challenges and solutions. PloS one, 14(8):e0221152. find
  14. Scheuerman, M. K., Jiang, J. A., Fiesler, C., and Brubaker, J. R. (2021). A framework of severity for harmful content online. Proc. ACM Hum.-Comput. Interact., 5(CSCW2):1–33. find
  15. Marusich, L., Bakdash, J. Z., Zhou, Y., and Kantarcioglu, M. (2024). Using AI uncertainty quantification to improve human decision making. In Proceedings of ICML, pp. 34949–34960. find
  16. Thorne, J., Vlachos, A., Christodoulopoulos, C., and Mittal, A. (2018). FEVER: A large scale dataset for fact extraction and verification. In Proceedings of NAACL HLT, pp. 809–819.
  17. Nakov, P., et al. (2021). Automated fact checking for assisting human fact checkers. In Proceedings of IJCAI, pp. 4551–4558. find
  18. Wang, W. Y. (2017). Liar, liar pants on fire: A new benchmark dataset for fake news detection. In Proceedings of ACL, pp. 422–426.
  19. Nakov, P., et al. (2022). Overview of the CLEF 2022 CheckThat! lab. In CLEF 2022, pp. 495–520. find
  20. Wang, S., et al. (2025). The unappreciated role of intent in algorithmic moderation. Harvard Kennedy School Misinformation Review, 6(3). find
  21. Nie, Y., Zhou, X., and Bansal, M. (2020). What can we learn from collective human opinions on natural language inference data? In Proceedings of EMNLP, pp. 9131–9143.
  22. Wardle, C. and Derakhshan, H. (2017). Information Disorder: Toward an Interdisciplinary Framework. Council of Europe Report DGI(2017)09. find
  23. Wong, K., Paritosh, P., and Aroyo, L. (2021). Cross-replication reliability: An empirical approach to interpreting inter-rater reliability. In Proceedings of ACL-IJCNLP, pp. 7053–7065. find
  24. Nielsen, D. S. and McConville, R. (2022). MuMiN: A large scale multilingual multimodal fact checked misinformation social network dataset. In Proceedings of SIGIR, pp. 3141–3153. find
  25. Xiao, Y., et al. (2022). Uncertainty quantification with pre trained language models: A large scale empirical analysis. In Findings of EMNLP, pp. 7273–7284. find
  26. Panchendrarajan, R. and Zubiaga, A. (2024). Claim detection for automated fact checking: A survey. Natural Language Processing Journal, 7:100066. find
  27. Xin, J., Tang, R., Yu, Y., and Lin, J. (2021). The art of abstention: Selective prediction and error regularization for NLP. In Proceedings of ACL, pp. 1040–1051. find
  28. Zaghouani, W., Jarrar, M., Habash, N., Bouamor, H., Zitouni, I., Diab, M., El-Beltagy, S. R., and AbuOdeh, M. (2024). The FIGNEWS shared task on news media narratives. arXiv preprint arXiv:2407.18147.
  29. Zhu, Y., et al. (2022). Generalizing to the future: Mitigating entity bias in fake news detection. In Proceedings of SIGIR, pp. 2120–2125. find
  30. Zubiaga, A., Aker, A., Bontcheva, K., Liakata, M., and Procter, R. (2018). Detection and resolution of rumours in social media: A survey. ACM Computing Surveys, 51(2):1–36. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/zaghouani-2026-grounding-information-disorder-nlp-theoretical/