Maximizing diversity within and among teams in a large-scale project

Abstract

This work aims to improve an earlier methodology for assigning personnel to diverse three-member teams. Notably, the original algorithm focused only on diversity within teams, to ensure that conflicting interests are represented in each team. While this indeed created diverse teams, in many cases different teams featured the same combination of conflicting interests. The client for the original project, a government agency, asked for a methodology that produces more combinations. Hence, the current study presents an approach for boosting diversity among teams. That is, the new method maximizes the differences among teams, not only differences within teams. This was achieved by limiting the number of times each combination appears, while maintaining maximal diversity within teams as well. The upgraded algorithm is scalable and fast converging.•We suggest an integer linear programming algorithm for creating thousands of diverse three-member teams that represent the conflicting interests of different groups.•The algorithm imposes penalty costs on potential assignments based on their deviations from the project's requirements and sets upper bound constraints on the frequency of different assignments.•The algorithm is efficient, scalable, and converges to maximal diversity within seconds.