Reviewer assignment for scientific articles using memetic algorithms

Abstract

In this work we modelled and solved the assignment problem appearing in MIC's paper review process using metaheuristic methods. Each given paper has to be reviewed by several different reviewers before being accepted for the conference. We implemented a memetic algorithm to solve that assignment problem and evaluated different model variants against their real world performance, using valuable feedback from many reviewers. While solutions generated by the solver alone already led to remarkable results compared to random solutions, making use of more expert knowledge throughout the solving process further improved solution quality. One way to achieve this was to fixate, prohibit or change solution parts manually and thus to iteratively build up a tuned solution. © 2007 by Springer Science+Business Media, LLC.