Paper 2509.21679
ReviewScore: Misinformed Peer Review Detection with Large Language Models
- Published
- Sep 2025
- Research lab
- Independent
- Citations
- 2
- GitHub
- 0 stars
01 In brief
Summary
The paper introduces REVIEWSCORE, a metric to detect misinformed review points in peer reviews, defined as questions answerable by the paper or weaknesses with incorrect premises.
Human annotation of ICLR reviews found 15.2% of weaknesses and 26.4% of questions misinformed.
The authors propose an automatic argument reconstruction engine that extracts explicit and implicit premises from weaknesses, using LLMs with feedback loops and a SAT solver to ensure validity and faithfulness.
They built a human expert-annotated dataset of 657 review points from 40 ICLR papers.
Evaluating eight LLMs, they found moderate human-model agreement (F1 0.4–0.5, kappa 0.3–0.4), with questions showing highest agreement and claims lowest.
Advanced REVIEWSCORE, based on premise-level factuality, outperformed base weakness-level scoring.
Disagreement analysis revealed models' incorrect reasoning and value judgments as main error sources.
Providing authors' responses improved argument evaluation but not question or claim scoring.
02 From the paper
Abstract
Peer review serves as a backbone of academic research, but in most AI conferences, the review quality is degrading as the number of submissions explodes. To reliably detect low-quality reviews, we define misinformed review points as either "weaknesses" in a review that contain incorrect premises, or "questions" in a review that can be already answered by the paper. We verify that 15.2% of weaknesses and 26.4% of questions are misinformed and introduce ReviewScore indicating if a review point is misinformed. To evaluate the factuality of each premise of weaknesses, we propose an automated engine that reconstructs every explicit and implicit premise from a weakness. We build a human expert-annotated ReviewScore dataset to check the ability of LLMs to automate ReviewScore evaluation. Then, we measure human-model agreements on ReviewScore using eight current state-of-the-art LLMs. The models show F1 scores of 0.4--0.5 and kappa scores of 0.3--0.4, indicating moderate agreement but also suggesting that fully automating the evaluation remains challenging. A thorough disagreement analysis reveals that most errors are due to models' incorrect reasoning. We also prove that evaluating premise-level factuality shows significantly higher agreements than evaluating weakness-level factuality.