DeepReview-13K
收藏资源简介:
DeepReview-13K是由浙江大学工程学院和西湖大学合作创建的一个结构化评审数据集。该数据集包含了原始研究论文、结构化的中间评审步骤和最终评估,旨在解决自动化评审系统中存在的缺乏细粒度专家评估过程的问题。DeepReview-13K数据集包含了13378条有效样本,是从两个ICLR会议周期(2024-2025)的18976篇论文投稿中收集并经过严格质量控制的。
DeepReview-13K is a structured review dataset co-created by the College of Engineering of Zhejiang University and Westlake University. This dataset encompasses original research papers, structured intermediate review procedures and final evaluations, aiming to address the critical gap where current automated review systems lack fine-grained expert evaluation processes. The DeepReview-13K dataset consists of 13,378 valid samples, which were collected from 18,976 paper submissions across two ICLR conference cycles (2024–2025) and subjected to strict quality control.

- 1DeepReview: Improving LLM-based Paper Review with Human-like Deep Thinking Process浙江大学工程学院,西湖大学 · 2025年



