Probability problems in knowledge acquisition for expert systems

The quality of an expert system, it is argued in this paper, is determined by the quality of its knowledge base. Many expert systems allow the use of uncertainty values. However, people have been found to be consistently susceptible to cognitive biases in estimating and manipulating probability values. Experts are no exception. This paper examines the factors which accentuate or modify these biases. The quality of any knowledge base which uses uncertainty values may be improved by attending to the problems outlined