Clustering Sequences of Complex Objects

All Contents

This paper is about the unsupervised discovery of patterns in sequences of composite objects. A composite object may be described as a sequence of other, simpler data. In such cases, not only the nature of the components is important, but also the order in which these components appear. The present work studies the problem of generalizing sequences of complex objects. A formal definition of generalized sequences is given, and an algorithm is derived. Because of the excessive computational complexity of this algorithm, a heuristic version is described. This algorithm is then integrated in a general-purpose clustering algorithm. The result is a knowledge discovery system which is able to analyze any structured database on the base of a unified, unsupervised mechanism.