Ensuring Data Security Against Knowledge Discovery in Distributed Information Systems

This paper contributes strategies that minimize the number of values which should be hidden in an Information System to guarantee that values of attributes containing sensitive information cannot be reconstructed by either Distributed or Local Chase within a Distributed Information System (DIS). The notion of the hidden attribute reconstruction by Knowledge Discovery and corresponding obstruction strategy was introduced in [1] where a minimal number of attribute values are additionally hidden to block the reconstruction of sensitive data by the rules extracted from DIS. The problem is particularly troublesome when locally generated rules restore those additionally hidden attribute values, and form a cycle of implications. The strategies in this paper complement existing Distributed Chase obstruction and work over widely used information systems.