(212a) The Automated NSF Panel Generator: An Elegant Optimal Solution to a Multi-Resource and Preference-Constrained Generalized Assignment Problem

This presentation is dedicated to the memory of Professor Christodoulos (Chris) A. Floudas and highlights a very important, but lesser-known, contribution of his research that led to the development of a mathematical model [1] and a software tool [2] for optimal assignment of reviewers to proposals being reviewed by a panel. With input from two NSF Program Directors, Maria Burka and T.J. Mountziaris, Chris Floudas and his students Stacy Janak and Martin Taylor formulated this problem as an integer linear programming problem that can be solved to optimality. They developed an elegant mathematical model for a multi-resource and preference-constrained assignment problem that optimizes the sum of a set of preference criteria for each reviewer while ensuring that each reviewer is assigned to approximately the same number of proposals and serves as "Lead Reviewer", "Scribe" or "Reviewer" with approximately the same frequency as other reviewers. Since its development in 2005, the panel assignment model and software tool have been used extensively by Program Directors at NSF (and elsewhere) to automatically generate more than 6,000 panels, thus saving the countless hours of personnel time which would otherwise be required to manually create and optimize these panels. The panel assignment algorithm has been adopted by NSF and is offered as an internal productivity tool to Program Directors.

References

1. Stacy L. Janak, Martin S. Taylor, Christodoulos A. Floudas, Maria Burka, and T. J. Mountziaris, Novel and Effective Integer Optimization Approach for the NSF Panel-Assignment Problem: A Multiresource and Preference-Constrained Generalized Assignment Problem, Ind. Eng. Chem. Res. 2006, 45, 1, 258–265.
2. Panel Assignment Problem Online Tool: http://ares.princeton.edu/casl_nsf_mgap/login.php