A comparison of the performance of some similarity and dissimilarity measures in the automatic classification of chemical structures

A group of 39 structures with local anesthetic activity has been classified automatically by calculating similarity or dissimilarity coefficients between pairs of structure diagrams and applying cluster analysis to the results. The performance of a number of similarity and dissimilarity coefficients has been compared using the relationship between structure and property. Simple coefficients and a distance function give more satisfactory results than functions using probabilistic weighting or standardized distance.