Analysis of Climate Campaigns on Social Media using Bayesian Model Averaging

Multiple stakeholders actively shape climate narratives on social media: The paper identifies that energy industries, their advocacy groups, and climate advocacy organizations all use Facebook advertising to influence public discourse on climate change, indicating this is not a one-sided conversation but rather a contested space where different interests compete for narrative control.

Automated detection of climate messaging is now possible: The researchers developed a machine learning approach using "model soup" and messaging themes that can automatically identify the stance (pro-climate action, pro-industry, etc.) of climate-related ads on Facebook, which could help researchers and the public better understand who is funding which messages.

Transparency tools for climate advocacy are being created: By releasing their stance dataset, trained model, and identified messaging themes publicly, the authors are providing resources that could enable future research into opinion mining and automatic detection of climate stances, potentially helping citizens identify the sources and intentions behind climate-related content they encounter online.

Social media campaigns respond to real-world events: The paper notes that climate advocacy campaigns on social media often emerge in response to ongoing societal concerns, suggesting that tracking these campaigns could reveal how different groups mobilize around climate-related developments in real time.