Combining social science and computer vision expertise, this project aims to understand which climate change-related images prevail on social media and what kind of reactions and emotions they elicit in debates among social media users.

Source: ClimateTV dataset (Prasse et al., 2023)

feature image

Source: Own visualization

Our research methods

We combine social media research with automated image analyses. Among others we use the following methods:

  • Image & Text Classification
  • Sentiment Analysis
  • Image Manipulation Detection
  • Image Clustering
  • Text Analysis
feature image

Source: Own visualization

Our data

We use social media from X, formely Twitter, for our analyses. To this end, we have downloaded material encompassing:

  • All content shared in the years 2019 - 2022.
  • English, German, French, Spanish, Arabic, Chinese, and Russian.
  • Image, video, text, user reactions, and user replies.
feature image

Source: Own visualization

Our toolbox

We will make the methods we employed easy accessible for other social science research project in creating a toolbox.

  • Comprehensive documentation of all methods.
  • Enable social sciences to employ automated methods during analyses.
  • Open source.
feature image

The ClimateVisions project is co-financed by the European Union and the Bundesministeriums für Bildung und Forschung (BMBF).