Dataset · tdiggelm
Climate change claim fact-checking dataset
tdiggelm/climate_fever
1,535 climate claims from the internet and 7,675 annotated evidence sentences from English Wikipedia.
What's inside
An English text corpus collected using the FEVER methodology. It includes 1,535 real internet climate claims, each matched with 5 sentences from English Wikipedia (7,675 pairs total). Manually annotated to determine whether the sentence supports, refutes, or provides insufficient information for the thesis.
What you can do with it
- Training and testing automated fact-checking and evidence retrieval systems
- Text classification for hypothesis verification or refutation (NLI)
- Testing RAG pipelines on complex ecology and climate queries
- Tasks
- Text classification, Text retrieval
- Languages
- English
- Size
- 1K–10K rows
- Downloads this month
- 2.4K
- Likes
- 34
Models trained on it
How to load it
from datasets import load_dataset
ds = load_dataset("tdiggelm/climate_fever")