ryokamoi/wice

This repository contains the dataset and code for "WiCE: Real-World Entailment for Claims in Wikipedia" in EMNLP 2023.

27
/ 100
Experimental

This project offers a comprehensive dataset of claims from Wikipedia, paired with supporting or refuting evidence from cited articles. It helps fact-checkers, researchers, and content validators analyze how well a claim is supported by its sources. You input a Wikipedia sentence and its cited articles, and it outputs whether the claim is fully supported, partially supported, or not supported, along with the specific supporting sentences or unsupported parts of the claim.

No commits in the last 6 months.

Use this if you need to rigorously evaluate natural language models designed for fact-checking, claim verification, or evidence retrieval in a real-world context using Wikipedia as a source.

Not ideal if you're looking for a simple, pre-trained model for immediate, out-of-the-box fact-checking without needing to train or evaluate on a detailed dataset.

fact-checking claim-verification evidence-analysis content-validation information-literacy
Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 8 / 25
Maturity 16 / 25
Community 3 / 25

How are scores calculated?

Stars

42

Forks

1

Language

Python

License

Last pushed

Dec 15, 2023

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/nlp/ryokamoi/wice"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.