On Assigning Papers to Reviewers

Meilong Huang, Bo Liu, Liang Hong

Abstract: We present a novel approach to conference paper assignment,i.e.,the task of assigning paper submissions to reviewers. In contradistinction to many proposed approaches, we allow for both preferences and expertise.In this paper,we seek to mine talent and useful information of reviewers'preference from the bidding result and to precisely estimate a reviewer's level of expertise for a given submitted paper. We explored four formulations to combine a reviewer's preference for a paper with the reviewer's expertise level for this paper and converted them into a single edge cost such that we can transform this problem into MCFP (the Minimum Cost of Flow Problem). Besides the average satisfaction, we proposed an"approximately stable"state to justify the assignment result.Finally we make a comparison between these formulations through simulation experiments and show that our formulations have great efficacy.