PeerConf: A dataset for peer review aggregation

PaperConf is a dataset that contains various crucial information about peer reviews from several scientific conferences. These Peer reviews are collected from 5 different scientific conferences with the permission of the conference chairs of the respective conferences. The dataset contains a total of 3242 peer reviews of 1236 papers from multiple scientific conferences. The peer reviews contain important information like recommendation score, confidence score, the review given by the reviewer, the final decision and some identifying attributes. Features like paper title, paper abstract, author name etc. as well as the conference names are removed during dataset preparation in order to maintain the privacy of the authors and reviewers. The dataset can be used for training and testing Machine Learning models predicting acceptance decisions, Peer review summarization etc.