Cooperative Personnel Scheduling (Research Project)

a joined research project from the Algorithms and Complexity Group, TU Wien, Austria, and Honda Research Institute, Germany

Project Team

Johannes Varga
Günther Raidl
Tobias Rodemann

Topic

Personnel scheduling tasks are common in the industry and often occur there on a regular basis. While the generated schedule directly affects the users, their preferences are normally not taken into account. We improve on this aspect by proposing a scheduling approach that respects the preferences of users as far as possible. Fully specifying all their preferences is tedious for the users. Therefore the users do not specify their preferences in the beginning but give feedback on schedules that are proposed by the system. These limited interactions are used to learn the user's preferences over time. The scheduling is done in a repeating fashion and the insights regarding user preferences of former iterations are used in subsequent iterations. To further improve the quality of the schedule while keeping user satisfaction high, user preferences may to some extent be influenced by problem-specific incentives, and the use of these incentives shall also be optimized.