GDBR: An Optimal Relation Generalization Algorithm for Knowledge Discovery from Databases

We present GDBR, Generalize DataBase Relation, an optimal, on-line O(n) algorithm for database relation generalization using concept hierarchies. The algorithm is a variant of attribute-oriented induction which generalizes database relations in either O(n log n) or O(np) time, where n is the number of input tuples and p is the number of tuples in the output relation. We augment the concept hierarchy structures to provide inherent generalization, define an encounter order on input concepts, progressively generalize input concepts as the need arises, and structure the final generalized relation in such a way to allow tuples to be inserted in one step.