COMPUTING DISSIMILARITIES BETWEEN OBJECTS: A BASIC STEP IN SYMBOLIC DATA ANALYSIS