QuotaClimat/frugalaichallenge-text-train

Dataset Card for Climate Disinformation quotes database from QuotaClimat & Data For Good

The dataset has been built for the Frugal AI Challenge 2025 co-located at the AI Action Summit in Paris and co-organized by Hugging Face & Data For Good. See more information at https://frugalaichallenge.org/

The goal of the Frugal AI Challenge is to encourage both academic and industry actors to keep efficiency in mind when deploying AI models. By tracking both energy consumption and performance for different AI for climate tasks, we can incentivize frugality in AI deployment while also addressing real-world challenges.

Dataset Summary

A comprehensive collection of approximately ~6500 climate-related quotes and statements, specifically focused on identifying and categorizing climate disinformation narratives. The dataset combines quotes and statements from various media sources, including television, radio, and online platforms, to help train models that can identify different types of climate disinformation claims. The labels are drawn from a simplified version of the CARDS taxonomy with only the 7 main labels.

@article{coan2021computer, title={Computer-assisted classification of contrarian claims about climate change}, author={Coan, Travis G and Boussalis, Constantine and Cook, John and others}, journal={Scientific Reports}, volume={11}, number={22320}, year={2021}, publisher={Nature Publishing Group}, doi={10.1038/s41598-021-01714-4} }

Coan, T.G., Boussalis, C., Cook, J. et al. Computer-assisted classification of contrarian claims about climate change. Sci Rep 11, 22320 (2021). https://doi.org/10.1038/s41598-021-01714-4

Dataset structure & labels

The dataset contains text data with associated labels representing different types of climate disinformation claims.

• text: The actual quote or claim about climate change
• label: Following categories:
  • 0_not_relevant: No relevant claim detected or claims that don't fit other categories
  • 1_not_happening: Claims denying the occurrence of global warming and its effects - Global warming is not happening. Climate change is NOT leading to melting ice (such as glaciers, sea ice, and permafrost), increased extreme weather, or rising sea levels. Cold weather also shows that climate change is not happening
  • 2_not_human: Claims denying human responsibility in climate change - Greenhouse gases from humans are not the causing climate change.
  • 3_not_bad: Claims minimizing or denying negative impacts of climate change - The impacts of climate change will not be bad and might even be beneficial.
  • 4_solutions_harmful_unnecessary: Claims against climate solutions - Climate solutions are harmful or unnecessary
  • 5_science_is_unreliable: Claims questioning climate science validity - Climate science is uncertain, unsound, unreliable, or biased.
  • 6_proponents_biased: Claims attacking climate scientists and activists - Climate scientists and proponents of climate action are alarmist, biased, wrong, hypocritical, corrupt, and/or politically motivated.
  • 7_fossil_fuels_needed: Claims promoting fossil fuel necessity - We need fossil fuels for economic growth, prosperity, and to maintain our standard of living.

{
    'text': 'Example claim about climate change...',
    'label': "1_not_happening",
}

Data Splits

The dataset is split into training and testing sets with a hidden test set for the challenge.

Dataset Creation

Curation & annoation

This dataset was compiled to help identify and understand common climate disinformation narratives in media and public discourse. It serves as a tool for training models that can automatically detect and categorize climate disinformation claims.

The dataset combines data from three main sources curated by the QuotaClimat & Data For Good team.

1. DeSmog climate disinformation database with extracted and annotated quotes with GPT4o-mini and manual validations
2. FLICC dataset from the paper "Detecting Fallacies in Climate Misinformation: A Technocognitive Approach to Identifying Misleading Argumentation " by Francisco Zanartu, John Cook, Markus Wagner, Julian Garcia - re-annotated with GPT4o-mini and manual validations
3. Quotes drawn from French media content (TV and radio) extracted by QuotaClimat & Data For Good using Mediatree tool, annotated manually.

Personal and Sensitive Information

The dataset contains publicly available statements and quotes. Care has been taken to focus on the claims themselves rather than personal information about individuals.

Getting Started

Example usage:

from datasets import load_dataset

dataset = load_dataset("quotaclimat/frugalaichallenge-text-train")
print(next(iter(dataset['train'])))

Additional Information

Licensing Information

The dataset is provided under the CC BY-NC 4.0 license.