Elamur Research⌕
☰
home / papers / 2026 / ClimateChat-300K: A Multi-Modal Facebook Dataset for Understanding Diverse Perspectives in Climate Communication

ClimateChat-300K: A Multi-Modal Facebook Dataset for Understanding Diverse Perspectives in Climate Communication

Summary

Studies of online climate discourse focus heavily on Twitter, partly because privacy rules and API limits make Facebook data hard to obtain. The authors address this gap with ClimateChat-300K. It contains 299,329 public English-language Facebook posts collected through CrowdTangle between May 2020 and May 2024, shortly before Meta shut down that tool.

Posts were retrieved with climate keywords covering general terms, policy and movement phrases, and activist names. They were then deduplicated and lightly cleaned, with hashtags and emojis kept. Each record has 41 page- and post-level fields, including text, timestamps, content type and detailed engagement counts.

The authors profile the corpus using several methods:

  • VADER-based polarity and subjectivity scores
  • quantile-based engagement tiers
  • chi-square association tests
  • topic modeling, which yields ten themes in five domains

They find that engagement levels vary with a post's sentiment, its media type and the page that published it. Visual and emotionally charged posts attract the most interaction, and posting activity rises around events such as COP summits and the pandemic period. Because CrowdTangle is no longer available, the dataset preserves Facebook climate discourse that would now be difficult to collect.

Key points

  • Scale and coverage: 299,329 posts with 41 features from 26,731 unique pages, collected 17 May 2020 to 16 May 2024. Activity averages about 205 posts per day and peaks at 1,572 posts on Earth Day 2021.
  • Content formats: links (142,513), photos (103,313) and native videos (38,236) dominate. Among reactions, Likes make up 71.9%, followed by Love (10.0%) and Haha (7.2%).
  • Sentiment profile: the vast majority of posts are classified as objective (about 97%). By polarity, 52.3% are positive, 26.9% negative and 20.8% neutral. Some counts in the text and tables do not match, so check the original before citing exact figures.
  • Sources: Sky News Australia leads with 435K interactions. US pages dominate, with smaller Global South representation from Kenya, Pakistan and the Philippines.
  • Access and limitations:
    • Released on Zenodo (record 18824386) under CC BY-NC-SA 4.0 for academic, non-commercial use.
    • Collection relied on keywords, which introduces selection bias.
    • Despite the "multi-modal" title, the data are English text plus metadata.
    • Collection cannot be replicated because CrowdTangle has closed.
    • Manual validation of the automatic labels was limited.

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: Linguistics & languages

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.63317/38nwpwivdq2x

@inproceedings{zaghouani2026climatechat,
  title = {{ClimateChat-300K: A Multi-Modal Facebook Dataset for Understanding Diverse Perspectives in Climate Communication}},
  author = {Zaghouani, Wajdi and Biswas, Md. Rafiul and Bessghaier, Mabrouka and Ibrahim, Shimaa Amer and Mikros, George},
  booktitle = {Proceedings of the Fifteenth Language Resources and Evaluation Conference},
  year = {2026},
  doi = {10.63317/38nwpwivdq2x},
  url = {https://doi.org/10.63317/38nwpwivdq2x}
}

References

18 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Christian Bachmaier. 2007. A radial adaptation of the sugiyama framework for visualizing hierarchical information. IEEE Transactions on Visualization and Computer Graphics, 13(3):583–594. find
  2. Ricardo Baeza-Yates. 2024. Introduction to responsible ai. In Proceedings of the 17th ACM International Conference on Web Search and Data Mining, pages 1114–1117. find
  3. Julia Bingler, Mathias Kraus, Markus Leippold, and Nicolas Webersinke. 2023. How cheap talk in climate disclosures relates to climate initiatives, corporate emissions, and reputation risk. Working paper, Available at SSRN 3998435. find
  4. Md. Rafiul Biswas and Wajdi Zaghouani. 2025. Enhancing Arabic dialectal sentiment analysis through advanced data augmentation techniques. In Proceedings of the Shared Task on Sentiment Analysis for Arabic Dialects, pages 24–28, Varna, Bulgaria. INCOMA Ltd., Shoumen, Bulgaria. find
  5. Meta Transparency Center. 2024. Crowdtangle. https://transparency.meta.com/he-il/researchtools/ other-datasets/crowdtangle/. Accessed: 2025-09-11. ↗
  6. Kirtika Deo and Abhnil Amtesh Prasad. 2020. Evidence of climate change engagement behaviour on a facebook fan-based page. Sustainability, 12(17):7038. find
  7. Max Falkenberg, Alessandro Galeazzi, Maddalena Torricelli, Niccolò Di Marco, Francesca Larosa, Madalina Sas, Amin Mekacher, Warren Pearce, Fabiana Zollo, Walter Quattrociocchi, et al. 2022. Growing polarization around climate change on social media. Nature Climate Change, 12(12):1114–1121. find
  8. Aleksandrina V Mavrodieva, Okky K Rachman, Vito B Harahap, and Rajib Shaw. 2019. Role of social media as a soft power tool in raising public awareness and engagement in addressing climate change. Climate, 7(10):122. find
  9. Niels G Mede and Ralph Schroeder. 2024. The “greta effect” on social media: A systematic review of research on thunberg’s impact on digital climate change communication. Environmental Communication, 18(6):801–818. find
  10. Angelina Mooseder, Cornelia Brantner, Rodrigo Zamith, and Jürgen Pfeffer. 2023. (social) media logics and visualizing climate change: 10 years of\# climatechange images on twitter. Social Media+ Society, 9(1):20563051231164310. find
  11. Sebastian Ruder, Anders Søgaard, and Ivan Vulić. 2019. Unsupervised cross-lingual representation learning. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics: Tutorial Abstracts, pages 31–38. find
  12. Alexandra Segerberg. 2017. Online and social media campaigns for climate change engagement. find
  13. In Oxford research encyclopedia of climate science. Oxford University Press. find
  14. Giuliana Sorce and Delia Dumitrica. 2023. From school strikes to webinars: Mapping the forced digitalization of fridays for future’s activism during the covid-19 pandemic. Convergence, 29(3):570– 585. find
  15. Roberto CSNP Souza, Denise EF de Brito, Rodrigo L Cardoso, Derick M de Oliveira, Wagner Meira Jr, and Gisele L Pappa. 2014. An evolutionary methodology for handling data scarcity and noise in monitoring real events from social media data. In Ibero-American Conference on Artificial Intelligence, page… find
  16. Saverio Storani, Max Falkenberg, Walter Quattrociocchi, and Matteo Cinelli. 2025. Relative engagement with sources of climate misinformation is growing across social media platforms. Scientific Reports, 15(1):18629. find
  17. Bebe Chand Sultana, Md Tabiur Rahman Prodhan, Edris Alam, Md Salman Sohel, ABM Mainul Bari, Subodh Chandra Pal, Md Kamrul Islam, and Abu Reza Md Towfiqul Islam. 2024. A systematic review of the nexus between climate change and social media: present status, trends, and future challenges. Frontiers in… find
  18. Quan Yuan, Gao Cong, Zongyang Ma, Aixin Sun, and Nadia Magnenat Thalmann. 2013. Who, where, when and what: discover spatio-temporal topics for twitter users. In Proceedings of the 19th ACM SIGKDD international conference on Knowledge discovery and data mining, pages 605–613. find

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/zaghouani-2026-climatechat-300k-multi-modal-facebook/