c2023-ripple-edits-wikidata-dependence

IN premise — summaries/2026/08/24/cohen-2023-ripple-effects-sA-acknowledgments.md

Created 2026-08-25T02:57:57+00:00

The RIPPLE EDITS benchmark relies on Wikidata as the external knowledge base for generating ripple-edit triplets, which may be incomplete or outdated

Summary

The RIPPLE EDITS benchmark builds its test questions by pulling from Wikidata, so any gaps or stale entries in Wikidata carry straight into the evaluation. That means a model could be scored unfairly — either penalized for not knowing something Wikidata itself misses, or tested against facts that have since changed.