ABSTRACT
The classification methods can be generally divided into two general classes: supervised classification and unsupervised classification. Partitioning methods aim at partitioning a set of objects into a specified number of disjoint groups in terms of a stated clustering criterion. Examples are given in Anderberg and Gordon. In Hierarchical methods, the structure of the data can be investigated at different levels. The dendrogram simultaneously summarizes the structure of the data at different levels. Clumps are overlapping groups. Overlapping of groups is allowed in this case. Such methods are applicable to word classification, since some words have different meanings. Geometric methods represent the data in a low-dimensional space and have been widely used in many disciplines. There are two methods of obtaining a geometric configuration of points whose interpoint distances approximate the dissimilarity between the corresponding pairs of objects. These are the principal coordinate analysis and the multidimensional sealing.
