Inference Control in Logic Databases as a Constraint Satisfaction Problem

We investigate inference control in logic databases. The administrator defines a confidentiality policy, i. e., the pieces of information which may not be disclosed to a certain user. We present a static approach which constructs an alternative database instance in which the confidential information is replaced by harmless information. The construction is performed by the means of constraint programming: The task of finding an appropriate database instance is delegated to a hierarchical constraint solver. We compare this static approach to a dynamic inference control mechanism - Controlled Query Evaluation - investigated in earlier work, and we also point out possible extensions which make use of the various opportunities offered by hierarchical constraint solvers.