Checking Classification Constraints for Consistency and Completeness

Abstract

Classification constraints are rules for assigning access classes to data when they are entered into a database. In order that a given set of constraints specify meaningful classes, they should be consistent, that is, not define conflicting classes for the same data; and complete, that is, assign a class to all data. This paper gives algorithms for checking the consistency and completeness of a set of classification constraints defined on a database schema. The techniques use computational geometry to compute intersecting regions in a multidimensional space, where each region is defined by a classification constraint or integrity constraint.