Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Augmented CARDS: A machine learning approach to identifying triggers of climate change misinformation on Twitter

Cristian Rojas, Frank Algra-Maschio, Mark Andrejevic, Travis Coan, John Cook, Yuan-Fang Li

TL;DR

The paper tackles the rapid spread of climate-m misinformation on social media by introducing Augmented CARDS, a two-stage hierarchical model that first performs binary detection (convinced vs contrarian) and then a multi-label contrarian taxonomy classification. Built on DeBERTa large, the approach is fine-tuned with platform-specific Twitter data and evaluated on both expert-annotated climate tweets ($n=2607$) and a large real-world Twitter corpus ($>5$ million tweets in 2022), showing substantial performance gains over the original CARDS model (e.g., binary F1 of $81.6$ and taxonomy F1 of $53.4$ on the expert set) and relative improvements of $16\%$ and $14.3\%$ respectively. The study reveals that over half of contrarian climate tweets attack climate actors or propagate conspiracy theories, and identifies four triggering categories for misinformation spikes: political events, natural events, contrarian influencers, and convinced influencers. This work supports near-real-time detection and potential automated responses, but acknowledges limitations such as English-language training, Twitter-centric data, and the need for multilingual and cross-platform extensions to generalize across information sources and cultures.

Abstract

Misinformation about climate change poses a significant threat to societal well-being, prompting the urgent need for effective mitigation strategies. However, the rapid proliferation of online misinformation on social media platforms outpaces the ability of fact-checkers to debunk false claims. Automated detection of climate change misinformation offers a promising solution. In this study, we address this gap by developing a two-step hierarchical model, the Augmented CARDS model, specifically designed for detecting contrarian climate claims on Twitter. Furthermore, we apply the Augmented CARDS model to five million climate-themed tweets over a six-month period in 2022. We find that over half of contrarian climate claims on Twitter involve attacks on climate actors or conspiracy theories. Spikes in climate contrarianism coincide with one of four stimuli: political events, natural events, contrarian influencers, or convinced influencers. Implications for automated responses to climate misinformation are discussed.

Augmented CARDS: A machine learning approach to identifying triggers of climate change misinformation on Twitter

TL;DR

The paper tackles the rapid spread of climate-m misinformation on social media by introducing Augmented CARDS, a two-stage hierarchical model that first performs binary detection (convinced vs contrarian) and then a multi-label contrarian taxonomy classification. Built on DeBERTa large, the approach is fine-tuned with platform-specific Twitter data and evaluated on both expert-annotated climate tweets () and a large real-world Twitter corpus ( million tweets in 2022), showing substantial performance gains over the original CARDS model (e.g., binary F1 of and taxonomy F1 of on the expert set) and relative improvements of and respectively. The study reveals that over half of contrarian climate tweets attack climate actors or propagate conspiracy theories, and identifies four triggering categories for misinformation spikes: political events, natural events, contrarian influencers, and convinced influencers. This work supports near-real-time detection and potential automated responses, but acknowledges limitations such as English-language training, Twitter-centric data, and the need for multilingual and cross-platform extensions to generalize across information sources and cultures.

Abstract

Misinformation about climate change poses a significant threat to societal well-being, prompting the urgent need for effective mitigation strategies. However, the rapid proliferation of online misinformation on social media platforms outpaces the ability of fact-checkers to debunk false claims. Automated detection of climate change misinformation offers a promising solution. In this study, we address this gap by developing a two-step hierarchical model, the Augmented CARDS model, specifically designed for detecting contrarian climate claims on Twitter. Furthermore, we apply the Augmented CARDS model to five million climate-themed tweets over a six-month period in 2022. We find that over half of contrarian climate claims on Twitter involve attacks on climate actors or conspiracy theories. Spikes in climate contrarianism coincide with one of four stimuli: political events, natural events, contrarian influencers, or convinced influencers. Implications for automated responses to climate misinformation are discussed.
Paper Structure (15 sections, 7 figures, 5 tables)

This paper contains 15 sections, 7 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Methods
  3. Model Architecture
  4. Training Details
  5. Data Analysis
  6. Results
  7. Assessing the Augmented CARDS model
  8. Applying Augmented CARDS to 2022 climate tweets
  9. Discussion
  10. Conclusion
  11. Data
  12. CARDS
  13. Waterloo
  14. Hamburg
  15. Anomalies: Word Analysis

Figures (7)

  • Figure 1: CARDS Taxonomy of Contrarian Climate Claims This taxonomy provides a comprehensive overview of the frequently employed main claim and its corresponding subarguments utilized to bolster contrarian perspectives on climate change.coan2021computer.
  • Figure 2: Model Architecture
  • Figure 3: Climate change Tweeter trends in 2022 (\ref{['fig:timeline']}) Number of tweets related to climate change topics by date in the 2022 period. (\ref{['fig:misinformation']}) Percentage of misinformative tweets detected by the CARDS and Augmented CARDS models.
  • Figure 4: Tweets sampled from the trend peak related to climate change observed between July 18 and July 21, 2022.
  • Figure 5: Tweets that deny the link between extreme weather and climate change.
  • ...and 2 more figures