Validating at early stages with a causal simulation tool

Abstract

Validating the dynamics of a conceptual model of expertise is a crucial task which is too often neglected and sometimes relegated after the implementation's phase. The motivation for this work is to capture and to simulate the dynamics of a modeled system during the early phases of design. In this paper, we present an approach and a tool based on a general and powerful simulation engine. We assume that the dynamics of a system can be viewed as a causal graph, where the nodes represent the parameters of the system and the links represent causal influences between these parameters. In a given state, the belief on the value of a parameter is an uncertain quantity represented by a probabilistic density over its domain of variation. We consider then semi-quantitative parameters and show that, using some results on discrete probabilities, we can exhibit a simulation method based on matrix calculus. We describe MORSE, a prototype based on a simple simulation algorithm, and illustrate its use on an example. Finally, we discuss the current limitations of this method and conclude about future developments of this work.