Causal Relational Learning
Babak Salimi, Harsh Parikh, Moe Kayali, Sudeepa Roy, Lise Getoor, Dan Suciu.
To be presented at SIGMOD 2020. Find the full paper here.
This is a dataset of academic peer-review. It consists of four semi-structured tables. The data covers two thousand submissions to ten conferences and workshops in computer science. The years 2017—2019 are represented. Importantly, it contains both accepted and rejected submissions.
ReviewData was created by compiling data from OpenReview, Scopus and the Shanghai University Rankings.
If you use this dataset, kindly cite the paper above.
Download (31 MB zipped, 390 MB unzipped)
papers.json: list of all submissions
paper_id: primary key
decision: whether the submission was accepted or rejected
title: name of the submission
authors: names of the authors
author_keys: unique IDs of the authors, foreign-key into
conf: conference submitted to, foreign-key into
- other fields include
authors.json: list of all authors with academic information
id: primary key
trueif only a single person found with that name (high confidence of match)
world-rank: ranking of institution, guessed via email domain or scopus name match (in that order).
name: full name
inst-guess: academic institution guess from email domain
scopus._json.coredata.document-count: lifetime papers published
scopus._json.coredata.citation-count: lifetime citations
scopus._json.coredata.coauthor-count: lifetime number of co-authors
scopus._json.coredata.author-profile.affiliation-current: current academic affiliation
scopus._json.coredata.author-profile.publication-range: first, last years of publishing
- many more fields available, such as
reviews.json: map from
paper_idto list of reviews for each submission
norm_conf: confidence of the reviewer, normalized to
norm_rating: reviewer’s rating of the submission, normalized to
title: review’s title
review: review’s text
rating: natural language rating of paper
confidence: natural language confidence of reviewer
confs.json: list of conferences
accept: how many submissions were accepted
reject: how many submissions were rejected
rigor: ratio of accept to reject