Reference: Gruber, T. R. & Gautier, P. Machine-generated Explanations of Engineering Models: A compositional modeling approach. Knowledge Systems Laboratory, Chambéry, France, 1993.
Abstract: We describe a method for generating causal explanations, in natural language, of the simulated behavior of physical devices. The method is implemented in DME, a system that helps formulate mathematical simulation models from a library of model fragments using a compositional modeling approach. Since the explanation system has access to the derivation of mathematical equations from the original model specification, the system can explain low-level quantitative behavior in terms of salient behavioral abstractions such as physical processes, idealized components, and operating modes. Several of the limitations of conventional explanation techniques are overcome. Instead of relying on ad hoc causal models, created specifically for the explanation task, the program infers causal relationships among parameters in the simulation model. Rather than using top-down sentence templates for text generation, DME composes textual annotations associated with model fragments into coherent sentences. We show how these and other explanation techniques are combined to provide informative explanations without imposing unrealistic demands on the model builder.
Notes: Updated July 1993.
Full paper available as ps.