Addressing time-scale differences among decision-makers through model abstractions

A decision support system (DSS) may, now more than ever, be shared by multiple decision-makers. Different decision-makers, however, can have different viewpoints and therefore different information needs in the use of a DSS. As a result, they may benefit from a system that is responsive to the special information needs of the individual user. This paper assumes, in particular, that decision-makers have varying requirements on the temporal dimension of the information. It also focuses on the model-base component of the DSS. Its underlying objective is automated model creation such that the model produced by the system is responsive to the distinct information needs of the user. I take a single, structural, dynamic model as a starting point, and develop a theoretical basis for the extension of a set of time scale-based abstractions of this model. The abstraction techniques, known as exogenization and equilibration, were originally proposed in qualitative reasoning (QR) research in artificial intelligence. I generalize the equilibration technique, and suggest a new time scale-based abstraction termed quasi-exogenization.