license: apache-2.0
task_categories:
- text-generation
language:
- en
size_categories:
- 100K<n<1M
1
Introduction
This dataset is the largest real-world consistency-ensured dataset for peer review, which features the widest range of conferences and the most complete review stages, including initial submissions, reviews, ratings and confidence, aspect ratings, rebuttals, discussions, score changes, meta-reviews, and final decisions.
Comparison with Existing Datasets
The comparison between our proposed dataset and existing peer review datasets is given below. Only the datasets in [bold] can guarantee all provided papers are initial submissions, which is critical for the consistency and quality of the review data. For the “Task” column, AP is the abbreviation for Acceptance Prediction, SP for Score Prediction, RA refers to Review Analysis tasks, RG stands for Review Generation, and MG denotes Meta-review Generation. In the three columns representing numbers, “-” means please refer to the original dataset for the number.
| Dataset Name | Data Source | Task | Reviews | Aspect Scores | Rebuttal | Score Changes | Meta-reviews | Final Decision | #Paper | #Review Comments | #Rebuttal |
|---|---|---|---|---|---|---|---|---|---|---|---|
| PeerRead | ICLR 17, ACL 17, NeurIPS 13-17 | AP, SP | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | 14,700 | 10,700 | 0 |
| DISAPERE | ICLR 19-20 | RA | ✓ | ✓ | ✓ | 0 | 506 | 506 | |||
| PEERAssist | ICLR 17-20 | AP | ✓ | ✓ | 4,467 | 13,401 | 0 | ||||
| [MReD] | ICLR 18-21 | MG | ✓ | ✓ | ✓ | ✓ | ✓ | 7,894 | 30,764 | 0 | |
| ASAP-Review | ICLR 17-20, NeurIPS 16-19 | RG | ✓ | ✓ | ✓ | ✓ | 8,877 | 28,119 | 0 | ||
| [NLPeer] | ACL 17, ARR 22, COLING 20, CONLL 16 | RA | ✓ | ✓ | ✓ | ✓ | 5,672 | 11,515 | 0 | ||
| PRRCA | ICLR 17-21 | MG | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | 7,627 | 25,316 | - |
| Zhang et al. | ICLR 17-22 | RA | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | 10,289 | 36,453 | 68,721 |
| [ARIES] | OpenReview | RG | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | 1,720 | 4,088 | 0 |
| AgentReview | ICLR 20-23 | RG, MG | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | 500 | 10,460 | - |
| Reviewer2 | ICLR 17-23, NeurIPS 16-22 (PeerRead & NLPeer) | RG | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | 27,805 | 99,727 | 0 |
| RR-MCQ | ICLR 17 (from PeerRead) | RG | ✓ | ✓ | 14 | 55 | 0 | ||||
| ReviewMT | ICLR 17-24 | RG | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | 26,841 | 92,017 | 0 |
| Review-5K | ICLR 24 | RG | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | 4,991 | 16,000 | 0 |
| DeepReview-13K | ICLR 24-25 | RG | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | 13,378 | 13,378 | 0 |
| [Our dataset] | 45 venues from 2017 to 2025 | RG, MG, AP, SP | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | 19,926 | 70,668 | 53,818 |