Application of Bayesian networks to the evaluation of roofing slate quality

We present the results of an application of Bayesian networks to the evaluation of the quality of roofing slate. Using data from borehole samples of a slate deposit, two networks constructed with different levels of expert knowledge input were evaluated for their capacities for inference and prediction of the quality of slate for roofing. be concluded from the results that Bayesian networks are an extremely useful automated tool for evaluating the quality of a resource such as slate for the following reasons: they allow final quality to be assessed immediately and in probabilistic terms with a tolerable degree of uncertainty; they enable the probability of obtaining different final qualities to be estimated when new data is introduced into the network; they speed up the evaluation process by simulating and optimising the work of the expert (during field data collection and borehole description) in identifying the parameters with the greatest influence on final quality; and finally, they have a satisfactory capacity for prediction.