Using activity theory and causal diagrams for designing multiagent systems that assist human activities

In this paper, we propose to use the Activity Theory and causal diagrams for modelling human activities with the aim of facilitating the specification of an agent-based assistance system through the Prometheus methodology. As a case study, we consider the elder medication activity, a recurring, complex and context-rich activity that involves several individual and collective activities which may require assistance. From the data collected in a contextual study of the elder medication, we modeled the medical consultation and refill medicine activities. Our results demonstrate that causal diagrams allow to capture the dynamics of the modelled activity, introduce the assistance of intelligent agents, extract the multiple scenarios synthesized in the activity structure and translate them into Prometheus artifacts.