CCF Database: a machine-learning-annotated corpus of Canadian climate articles
Presented at EPSS (Belfast, June 18-20, 2026), this presentation introduces the CCF Database: a corpus of 266,271 Canadian climate articles from 20 newspapers, annotated at the sentence level across 65 categories. The talk focuses on why framing choices matter, how the annotation schema organizes competing climate narratives, and how the database supports transparent validation and comparative media analysis.
A corpus built for comparing climate frames
The presentation shows how the same events can be framed differently across outlets, regions, and moments, using the 65-category annotation schema to distinguish scientific, political, economic, social, and responsibility-oriented interpretations of climate change.
Validation is part of the database design
The deck presents the validation logic behind the corpus, including category organization, model performance checks, and public documentation that make the annotation choices inspectable and reusable.
Scroll Navigate between slides Click Interact with data Fullscreen for a better experience
Use arrow keys or click to navigate through the slides.