Elamur Research⌕
☰
home / papers / 2026 / Retrieving Climate Change Disinformation by Narrative

Retrieving Climate Change Disinformation by Narrative

Paper · 2026 · Max Upravitelev, Veronika Solopova, Charlott Jakob et al.

Summary

Climate disinformation narratives are usually detected by classifiers built on fixed taxonomies, so new storylines go unnoticed. The authors instead treat narrative detection as retrieval. An analyst writes a narrative's core message as a query, and the system ranks corpus texts by how well they fit it, with no predefined label set. This is hard because narrative descriptions are abstract, while real texts are concrete, varied and often only imply the message.

The authors turn three datasets (CARDS, Climate Obstruction and the climate subset of PolyNarrative) into retrieval benchmarks and propose SpecFi. Following HyDE, SpecFi asks an LLM to write hypothetical disinformation texts for each narrative and then uses them for dense retrieval. Few-shot examples come either from the nearest unlabeled training text (SpecFi-DR) or from NodeRAG community summaries (SpecFi-CS). SpecFi-CS reaches 0.494 MAP on CARDS without narrative labels. A new difficulty measure, narrative variance, shows that standard retrievers degrade on diverse narratives, while SpecFi-CS stays comparatively stable. This matters for fact-checkers who need to track emerging narratives.

Key points

  • Pipeline: The training split serves as an unlabeled reference corpus and the test split is evaluated. Queries are built by joining hierarchical taxonomy labels, and n=10 hypothetical documents are generated per narrative. Their embeddings are aggregated for dense retrieval.
  • SpecFi-CS: This variant builds a heterogeneous NodeRAG graph with seven node types and applies Leiden community detection to produce LLM-written community summaries. It then retrieves the best summary through Personalized PageRank search. These summaries worked better as few-shot examples than real training texts.
  • Narrative variance (V_i): This embedding-based measure captures how spread out the texts of a narrative are. Partial correlation analysis controlling for sample size links it to retrieval difficulty. On high-variance narratives, BM25 loses 63.4% of its MAP, whereas SpecFi-CS loses 32.7%.
  • Summaries vs. expert taxonomy: Unsupervised community summaries often match the expert CARDS taxonomy. Failure cases fall into four types:
    • partial alignment;
    • collapse, where sibling sub-narratives share one summary;
    • drift;
    • incoherent output.
  • Code: Released at github.com/XplaiNLP/SpecFi-Narrative-Retrieval (TU Berlin / DFKI; NLP4Ecology @ LREC 2026).

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

🔬German Research Centre for Artificial IntelligenceKaiserslautern · Germany77 works · 29,479 citations🎓Technische Universität BerlinBerlin · Germany22 works · 957 citations

 Fields, topics and concepts

📚 Shelves 7The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 worksSocial Sciences · 1,772 worksInformation Systems · 1,437 worksSociology and Political Science · 577 works

Cite

Cite as Max Upravitelev, Veronika Solopova, Charlott Jakob, Premtim Sahitaj, Sebastian Möller and Vera Schmitt. 2026. Retrieving Climate Change Disinformation by Narrative. Proceedings of the 2nd Workshop on Ecology, Environment, and Natural Language Processing.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.63317/5markfecdiyu

@inproceedings{upravitelev2026retrieving,
  title = {{Retrieving Climate Change Disinformation by Narrative}},
  author = {Max Upravitelev, Veronika Solopova, Charlott Jakob, Premtim Sahitaj, Sebastian Möller and Vera Schmitt},
  booktitle = {Proceedings of the 2nd Workshop on Ecology, Environment, and Natural Language Processing},
  year = {2026},
  doi = {10.63317/5markfecdiyu},
  url = {https://doi.org/10.63317/5markfecdiyu}
}

References

32 references, 5 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Mousumi Akter and Shubhra Kanti Karmaker Santu. 2024. Fans: a facet-based narrative similarity metric. find
  2. Darren Edge, Ha Trinh, Newman Cheng, Joshua Bradley, Alex Chao, Apurva Mody, Steven Truitt, Dasha Metropolitansky, Robert Osazuwa Ness, and Jonathan Larson. 2025. From local to global: https://terry-xu-666.github.io/NodeRAG_web/blog/2025/03/16/ structure-output/ A graph rag approach to query-focuse… ↗
  3. Guillaume Jacquet, Vlad Dragu, Leonida della Rocca, Stefano Bucci, Aldo Podavini, Marco Verile, Charles Macmillan, and Jens P. Linge. 2023. Trend analysis of COVID-19 mis/disinformation narratives–A 3-year study. PLOS ONE, 18(11):e0291423. Publisher: Public Library of Science. find
  4. Luyu Gao, Xueguang Ma, Jimmy Lin, and Jamie Callan. 2023. Precise zero-shot dense retrieval without relevance labels. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 1762–1777, Toronto, Canada. Association for Computational Li…
  5. Zehan Li, Xin Zhang, Yanzhao Zhang, Dingkun Long, Pengjun Xie, and Meishan Zhang. 2023. Towards general text embeddings with multi-stage contrastive learning.
  6. Saibo Geng, Hudson Cooper, Michał Moskal, Samuel Jenkins, Julian Berman, Nathan Ranchin, Robert West, Eric Horvitz, and Harsha Nori. 2025. Jsonschemabench: A rigorous benchmark of structured outputs for language models.
  7. Christopher D. Manning, Prabhakar Raghavan, and Hinrich Schütze. 2008. Introduction to Information Retrieval. Cambridge University Press. find
  8. OpenAI. 2024. Gpt-4o system card.
  9. Fatima Haouari, Carolina Scarton, Nicolò Faggiani, Nikolaos Nikolaidis, Bonka Kotseva, Ibrahim Abu Farha, Jens Linge, and Kalina Bontcheva. 2025. UKElectionNarratives: A Dataset of Misleading Narratives Surrounding Recent UK General Elections. Proceedings of the International AAAI Conference on Web… find
  10. Andrew Piper, Richard Jean So, and David Bamman. 2021. Narrative theory for computational narrative understanding. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 298–311, Online and Punta Cana, Dominican Republic. Association for Computational Lingui… find
  11. Hanna-Riikka Roine. 2020. On speculation as a strategy. Fafnir – Nordic Journal of Science Fiction and Fantasy Research, 7(2):8–15. find
  12. Hans Ole Hatzel and Chris Biemann. 2024. Story embeddings — narrative-focused representations of fictional stories. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 5931–5943, Miami, Florida, USA. Association for Computational Linguistics. find
  13. Ariana Sahitaj, Premtim Sahitaj, Veronika Solopova, Jiaao Li, Sebastian Möller, and Vera Schmitt. 2025. Hybrid annotation for propaganda detection: Integrating LLM pre-annotations with human intelligence. In Proceedings of the Fourth Workshop on NLP for Positive Impact (NLP4PI), pages 215–228, Vienn… find
  14. Philipp Heinrich, Andreas Blombach, Bao Minh Doan Dang, Leonardo Zilio, Linda Havenstein, Nathan Dykes, Stephanie Evert, and Fabian Schäfer. 2024. Automatic identification of COVID-19-related conspiracy narratives in German telegram channels and chats. In Proceedings of the 2024 Joint International… find
  15. Yanzhen Shen, Sihao Chen, Xueqiang Xu, Yunyi Zhang, Chaitanya Malaviya, and Dan Roth. 2025. Logicol: Logically-informed contrastive learning for set-based dense retrieval. find
  16. Iknoor Singh. 2024. Detecting and Tracking the Spread of Debunked Narratives Across Languages. Phd thesis, University of Sheffield, Sheffield, UK. Supervisors: Carolina Scarton and Kalina Bontcheva. find
  17. Maria Hellman. 2024. Narrative Analysis and Framing Analysis of Disinformation, pages 101–121. Springer Nature Switzerland, Cham. find
  18. Iknoor Singh, Carolina Scarton, Xingyi Song, and Kalina Bontcheva. 2024. Breaking language barriers with mmtweets: Advancing cross-lingual debunked narrative retrieval for fact-checking. find
  19. Arman Irani, Ju Yeon Park, Kevin Esterling, and Michalis Faloutsos. 2025. A discourse analysis framework for legislative and social media debates. In Proceedings of the 17th ACM Web Science Conference 2025, Websci ’25, page 199–209, New York, NY, USA. Association for Computing Machinery. find
  20. Veronika Solopova, Christoph Benzmüller, and Tim Landgraf. 2023. The evolution of pro-kremlin propaganda from a machine learning and linguistics perspective. In Proceedings of the Second Ukrainian Natural Language Processing Workshop (UNLP), pages 40–48, Dubrovnik, Croatia. Association for Computati… find
  21. Bonka Kotseva, Irene Vianini, Nikolaos Nikolaidis, Nicolò Faggiani, Kristina Potapova, Caroline Gasparro, Yaniv Steiner, Jessica Scornavacche, 10. find
  22. Witold Sosnowski, Arkadiusz Modzelewski, Kinga Skorupska, Jahna Otterbacher, and Adam Wierzbicki. 2024. EU DisinfoTest: a benchmark for evaluating language models’ ability to detect disinformation narratives. In Findings of the Association for Computational Linguistics: EMNLP 2024, pages 14702–14723… find
  23. Language Resource References Coan, Travis G. and Boussalis, Constantine and Cook, John and Nanko, Mirjam O. 2021. CARDS: Computer-Assisted Recognition of Denial and Skepticism – Climate Change Contrarian Claims Dataset. distributed with the article in Scientific Reports. PID https://doi.org/10.1038/… — [doi:10.1038/s41598-021-01714-4](https://doi.org/10.1038/s41598-021-01714-4) find
  24. Gemma Team, Aishwarya Kamath, Johan Ferret, Shreya Pathak, et al. 2025. Gemma 3 technical report. find
  25. V. A. Traag, L. Waltman, and N. J. van Eck. 2019. From Louvain to Leiden: guaranteeing well-connected communities. Scientific Reports, 9(1):5233. find
  26. Nikolaidis, Nikolaos and Stefanovitch, Nicolas and Silvano, Purificação and Dimitrov, Dimitar Iliyanov and Yangarber, Roman and Guimarães, Nuno and Sartori, Elisa and Androutsopoulos, Ion and Nakov, Preslav and Da San Martino, Giovanni and Piskorski, Jakub. 2025. PolyNarrative: A Multilingual, Multi… find
  27. Ivan Vykopal, Matúš Pikuliak, Ivan Srba, Robert Moro, Dominik Macko, and Maria Bielikova. 2024. Disinformation capabilities of large language models. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 14830– 14847, Bangkok, Thail… find
  28. Tianyang Xu, Haojie Zheng, Chengze Li, Haoxiang Chen, Yixin Liu, Ruoxi Chen, and Lichao Sun. 2025. Noderag: Structuring graph-based rag with heterogeneous nodes. find
  29. Rowlands, Harri and Morio, Gaku and Tanner, Dylan and Manning, Christopher. 2024. Climate Obstruction in Social Media Advertising Dataset. Association for Computational Linguistics. PID https://doi.org/10.18653/v1/2024.findings-acl.330. Dataset of social media advertisements from oil and gas compani… ↗
  30. An Yang, Anfeng Li, Baosong Yang, Beichen Zhang, et al. 2025. Qwen3 technical report.
  31. Yan Zhang, Zhaopeng Feng, Zhiyang Teng, Zuozhu Liu, and Haizhou Li. 2023. How well do text embedding models understand syntax? In Findings of the Association for Computational Linguistics: EMNLP 2023, pages 9717–9728, Singapore. Association for Computational Linguistics. find
  32. Lixing Zhu, Runcong Zhao, Lin Gui, and Yulan He. 2023. Are NLP models good at tracing thoughts: An overview of narrative understanding. In Findings of the Association for Computational Linguistics: EMNLP 2023, pages 10098–10121, Singapore. Association for Computational Linguistics. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/upravitelev-2026-retrieving-climate-change-disinformation-narrative/