Identify Reviewers from their Comments
Abstract
The peer-review process is quintessential for ensuring the validity of scientific research and provides assessment about the importance of scientific publications. By definition, the peer-review process should be critical and rigorous. To guarantee this, it is important to ensure reviewers can express their opinions without the concerns of targeted harassment from the authors. Therefore, in most peer review process, reviewers are anonymous. Yet, reviewer comments unavoidably provide some information about the identities of the reviewers. In our work, we showed that once we can narrow down the reviewers to a reasonable amount of candidates, simple author attribution algorithms trained on the public-available corpus from the candidates allow identifying the reviewers with relatively high accuracy. Our work has important implications for privacy protection during and after the peer-review process.