AgentReview: Exploring Peer Review Dynamics with LLM Agents

Peer review is fundamental to the integrity and advancement of scientificpublication. Traditional methods of peer review analyses often rely onexploration and statistics of existing peer review data, which do notadequately address the multivariate nature of the process, account for thelatent variables, and are further constrained by privacy concerns due to thesensitive nature of the data. We introduce AgentReview, the first largelanguage model (LLM) based peer review simulation framework, which effectivelydisentangles the impacts of multiple latent factors and addresses the privacyissue. Our study reveals significant insights, including a notable 37.1

Further reading#

• Access Paper in arXiv.org