DeepActor Models in DANUBIA

Abstract

This chapter describes the representation of decision processes of socio-economic actors by means of actor models, which is one distinctive feature of DANUBIA. An actor model (also called an agent model) describes socio-economic processes as the sum of the individual actions taken by a range of different actors. The DeepActor framework provides a basis for modelling and implementing the socio-economic DANUBIA models. Decision makers, such as individuals, organisations or businesses, are modelled as “actors”. Each actor is localised within his physical and social environment (a proxel or a social network) and takes decisions as responses to his observations about which action to execute from a range of possible action options. Actors have various preferences and action options, represented by customised plans and decision procedures that are specific to each actor type. Further, actors have a “memory” (history) for recalling previous decisions. One special feature of DANUBIA is the coupling of physically based scientific models with socio-economic components. It is described how the transformation from quantitative states in nature to qualitative notifications for the actor model is realised using the flag concept.