Bad practices in open peer review: lessons learned from mining MDPI open reports

The reviewing process is a key step in the publication of scientific papers. Unreliable scientific papers exist, some leading to retractions. The publication of such flawed papers reveals flaws in the peer review process, which is generally not made public. However, some rare publishers such as MDPI provide access to some of the reviewers’ reports on articles accepted in its journals. This talk presents various examples of questionable practices in the review process, revealed by applying text mining techniques to a corpus of open review reports of MDPI journals from 2011 to 2022. These range from extremely short reports to identical reports, including the reuse of text passages in several reports. These findings about disclosed reports cannot but raise questions about the trustworthiness of the vast majority of reports that are not made public.