A Case for Numerical Taxonomy in Case-Based Reasoning

There are applications of case-like knowledge where, on the one hand, no obvious best way to structure the material exists, and on the other, the number of cases is not large enough for machine learning to find regularities that can be used for structuring. Numerical taxonomy is proposed as a technique for determining degrees of similarity between cases under these conditions. Its effect is illustrated in a novel application for case-like knowledge: authentication of paintings.