Fair Reviewer Assignment Considering Academic Social Network

An important task in peer review is assigning papers to appropriate reviewers, which is known as Reviewer Assignment (RA) problem. Most of existing works mainly focus on the similarity between paper topics and reviewer expertise, few works consider multiple relationships between authors and reviewers on academic social network. However, these relationships could influence reviewer assessments on fairness. In this paper, we address RA problem considering academic social network to find a confident, fair and balanced assignment. We model papers and reviewers based on matching degree by combining collaboration distance and topic similarity, and propose Maximum Sum of Matching degree RA (MSMRA) problem. Two algorithms are designed for MSMRA problem: Simulated Annealing-based Stochastic Approximation, and Maximum Matching and Minimum Deviation. Experiments show that our methods achieved good performance both on overall effectiveness and fairness distribution within reasonable running time.