Relational Learning with Decision Trees.
Peter Geibel, Fritz Wysotzki · 1996
. In this paper, we describe two different learning tasks for relational structures. When learning a classifier for structures, the relational structures in the training sets are classified as a whole. Contrarily, when learning a context dependent classifier for elementary objects, the elementary objects of the relational structures in the training set are classified. We investigate the question how such classifications can be induced automatically from a given training set containing classified structures or classified elementary objects respectively. We present an algorithm based on fast graph isomorphism testing that allows the description of the objects in the training set by automatically constructed attributes. This allows us to employ wellknown methods of decision tree induction to construct a hypothesis. We describe new simplification and structure reconstruction techniques for the learned structural decision tree. We present the system INDIGO and evaluate it on the Mesh and th...