CoCoReviewBench curates 3,900 conference papers with category subsets and expert discussion annotations to evaluate AI reviewers on completeness and correctness, showing they are limited and prone to hallucinations while reasoning models perform better.
- If the candidate is much more generic than a gold opinion, that gold opinion should be treated as *not covered * or only weakly covered
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
CoCoReviewBench: A Completeness- and Correctness-Oriented Benchmark for AI Reviewers
CoCoReviewBench curates 3,900 conference papers with category subsets and expert discussion annotations to evaluate AI reviewers on completeness and correctness, showing they are limited and prone to hallucinations while reasoning models perform better.