Reducing Subjectivity in the System Dynamics Modeling Process: An Interdisciplinary Approach

摘要

Over the last six decades, the system dynamics (SD) methodology has been used to address a variety of issues that are inherent to complex systems. In the SD methodology, circular-causality-centric problem definitions and computer simulations are uniquely applied to narrow the scope of a problem and facilitate an understanding of the diverse phenomena that arise from the underlying problem structure. With unclear but conceivable causality between the variables that constitute a problem, intentionally choosing to focus on the circular causality (causal loop) may result in erroneous models. The SD modeling tools, causal loop diagram for mental modeling, and stocks-and-flows diagram for physical simulation modeling are all interconnected, and it is therefore important to maintain consistency between their outputs to ensure the procedural validity. However, the modeling activities are normally performed individually, which introduces ambiguity and subjectivity into the SD modeling process. To address this issue, this research presents an integrated SD modeling support system that employs graph theory, text mining, and social network analysis approaches, and can be used as an extended SD modeling framework. This system is expected to facilitate SD modeling and to reduce errors in the SD modeling process. In the future, it will be utilized to implement an advanced computer-aided SD modeling toolchain and methodology.