Towards a human-centered decision support system for scheduling with shifts

The production supervisor job is known to be highly stressful, requiring humans to react quickly to complex and unforeseen socio-organizational situations, often under high time and hierarchical pressure, where inappropriate reactions can have disastrous economic consequences. In such situations, decision support systems (DSS) can be providential to help decision-makers make the best of their decisions, provided that such systems meet the desired levels of usability, acceptability, and effectiveness. Focusing on a particular industrial use case, this paper presents the outcomes of a design process centered on human needs, that keeps supervisors at the heart of the decision-making. We detail the chosen DSS architecture, the essence of the decision problems faced by supervisors, as well as a constraint programming (CP) approach able to deal with them. The paper also shows how CP can be coupled with human-computer interactions, and proposes an ecological serious game that enables to experimentally assess the usability, acceptability, and effectiveness of the DSS.