Abstract: | The task of generating informative explanations in industrial training involves automated formulation of system models with respect to the varying levels of the trainees' knowledge. Compositional Modelling provides a useful basis upon which to structure a suite of models that may reflect different complexities of the systems being modelled. However, additional inferences are required in order to select appropriate model fragments to form a coherent system model that is suitable for a given trainee's degree of expertise. This paper presents a novel approach to perform such inferences by the use of Bayesian networks. The work is implemented and typical experimental results are given.
|