In philosophy of science, the interest for computational models and simulations has increased heavily during the past decades. Different positions regarding the validity of models have emerged but the views have not succeeded in capturing the diversity of validation methods. The wide variety of models with regards to their purpose, character, field of application and time dimension inherently calls for a similar diversity in validation approaches. A classification of models in terms of the mentioned elements is presented and used to shed light on possible types of validation leading to a categorisation of validation methods. Through different examples it is shown that the methods of validation depend on a number of things as the context of the model, the mathematical nature, the data available and the representational power of the model. In philosophy of science many discussions of validation of models has been somewhat narrow-minded reducing the notion of validation to establishment of truth. This article puts forward the diversity in applications of simulation models that demands a corresponding diversity in the notion of validation.
|Tidsskrift||Journal of Experimental & Theoretical Artificial Intelligence|
|Status||Udgivet - 2012|