ARData: A Comprehensive Dataset for Reviewer Selection and Expertise Matching for Article Evaluation

This is a carefully curated dataset aimed at streamlining the reviewer selection process to ensure high-quality evaluations of submitted manuscripts, called Article Review Dataset (ARData). It was initially developed for scholarly conferences but is also adaptable for journal reviewer selection. The dataset is structured around three core components: Reviewer Data (RD), Manuscript Data (MD), and Reviewer Confidence Scores (RCS).

The RD contains keyphrases extracted from reviewers' publications over the past decade. The MD includes keyphrases derived from submitted manuscripts. The RCS offers a quantitative measure of reviewer confidence in evaluating manuscripts, providing a metric to assess the review process.

To maintain privacy, reviewer identities are anonymized using the SHA1 algorithm, with only the least significant 8-byte segment of the 20-byte hash values retained to optimize processing speed. This dataset is a valuable tool for testing and validating recommender systems, with applications in reviewer selection for conferences and journals, evaluator selection for grants, expert team formation, and other related tasks.