Detecting Bias in Climate Reporting in English and German Language News Media

Challenge Background

Beyond natural climate variability, human-induced climate change has had extensive negative effects on nature and people, including more frequent and powerful extreme events, according to the United Nations’ Intergovernmental Panel on Climate Change (IPCC). The reported increase in the frequency and intensity of climate and weather extremes, such as hot extremes on land and in the ocean, heavy precipitation events, drought, and fire weather, has had widespread, pervasive effects on ecosystems, people, towns, and infrastructure.

Meanwhile, media coverage of the climate crisis has only recently improved in engagement and accuracy. This improvement is severely weighed down by scientifically misleading, deliberately polarizing and fake information meted out through media outlets with political or ideological biases. IPCC’s Sixth Assessment Report acknowledges that “Accurate transference of the climate science has been undermined significantly by climate change countermovements, in both legacy and new/social media environments through misinformation”, negatively impacting climate governance. The outcome is stalled mitigation of the climate emergency in many countries where its health impacts can be deadlier than cancer. Research has shown Climate reporting in the Global North has significant regional bias and a tendency to report as ‘detached observers’, giving space to debates around the facts of climate science and ignoring climate-related humanitarian emergencies in the Global South and indigenous populations. Misinformation and trolling campaigns motivated by biased political and fossil fuel-dependent organizations are also a major challenge in this regard.

The Problem

We will develop a user-friendly bias-rating system to rate the accuracy of climate-related reports in comparison with established facts in climate science and related areas. For this, we will analyze English and German language news media using Natural Language Processing.

Goal of the Project

The overarching goals of the system would be:

• Check the content of written articles against credible sources of climate science
• Calculate a bias score for each article 
• Highlight scientifically inaccurate assertions in written articles 
• Provide the credible source to the user in an easy-to-access format (e.g. through a Streamlit dashboard/app/website)

A credible source of information can be in any form, from satellite imagery of areas affected by climate-related incidents to graphical information in scientific papers and reports. Therefore, this object will also include elements of image recognition, data analysis and data visualization.

Project Timeline

What you'll learn

1. Natural Language Processing
2. Data Visualization and Analysis
3. Dashboard Representation
4. Web Scraping and Labelling