Keywords
ontology, integrated assessment, model evaluation, conceptual model
Start Date
1-7-2010 12:00 AM
Abstract
Integrated assessment models are used to analyze global change issues and they allow a better understanding of our complex environment. It is crucial to be able to relate these models to their scientific basis, both for interpretation and validation purposes. Current model evaluation procedures focus on model behavior analysis; the conceptual knowledge and assumptions embedded in integrated assessment models are hardly tested. As such, current model evaluation procedures do not contribute to the understanding of the structure of the models and the selected mechanisms and assumptions. We submit that evaluation of the scientific basis of integrated assessment models should follow the standard procedures for evaluation of scientific theories, which implies that these models should be subjected to critical peer reviews. However, much knowledge is hidden in the source code and therefore not accessible to peers. In this paper we propose to use ontologies – explicit specifications of shared conceptual knowledge – to represent the knowledge encoded in integrated assessment models in a clear and transparent way. We show the proposed peer review evaluation procedure in a case study concerning a system-dynamics model on residential energy use in India. We found that the ontology helped peers to obtain more information on the model and to gain more insight in its structure. However, a better balance between different types of model documentation and explicit links between them are needed to improve the understanding by the peers. We believe that ontologies can be exploited further in a computational sense in order to achieve model transparency.
The use of ontologies in peer reviews of Integrated Assessment Models
Integrated assessment models are used to analyze global change issues and they allow a better understanding of our complex environment. It is crucial to be able to relate these models to their scientific basis, both for interpretation and validation purposes. Current model evaluation procedures focus on model behavior analysis; the conceptual knowledge and assumptions embedded in integrated assessment models are hardly tested. As such, current model evaluation procedures do not contribute to the understanding of the structure of the models and the selected mechanisms and assumptions. We submit that evaluation of the scientific basis of integrated assessment models should follow the standard procedures for evaluation of scientific theories, which implies that these models should be subjected to critical peer reviews. However, much knowledge is hidden in the source code and therefore not accessible to peers. In this paper we propose to use ontologies – explicit specifications of shared conceptual knowledge – to represent the knowledge encoded in integrated assessment models in a clear and transparent way. We show the proposed peer review evaluation procedure in a case study concerning a system-dynamics model on residential energy use in India. We found that the ontology helped peers to obtain more information on the model and to gain more insight in its structure. However, a better balance between different types of model documentation and explicit links between them are needed to improve the understanding by the peers. We believe that ontologies can be exploited further in a computational sense in order to achieve model transparency.
