Team formation in business process context

Abstract

High performing teams may benefit the execution efficiency of business processes. In executing a process, the performance of the team is concerned with the individual expertise of team members and the handover relations between executors of adjacent tasks of the process. Team formation problem in the presence of business process can be defined as finding a group of individuals to execute all tasks of the process. Considering individual expertise and handover relations, a method called Bayesian Network based Team Formation (BN-TF) is proposed to address the team formation problem in business process context. Using BN-TF, a team formation problem is first modeled as a Most Probable Explanation (MPE) problem in Bayesian network based on the structural information of the related business process. For solving the transformed MPE problem, we design an improved genetic algorithm called Forward-Backward Greedy Genetic Algorithm (FBG-GA). Experimental results on simulation data verify that BN-TF indeed produces teams that satisfy the requirements while improve the execution efficiency of the business process. Compared with existing methods, RarestFirst and CoverSteiner, which focus on minimizing team communication cost, BN-TF shows improvement in terms of individual expertise and handover relations.