Consistency and Plausible Inference

The usual approach to plausible reasoning is to associate a validity measure with each fact or rule, and to compute from these a validity measure for any deduction that is made. This approach is shown to be inappropriate for some classes of problems, particularly those in which the evidence is not internally consistent. The shortcomings of two current plausible reasoning architectures are demonstrated on one such program. The paper then outlines a new approach based on the discovery of consistent subsets of the given evidence and presents results from an implementation of this scheme.