Inference Based Security Database Auditing Framework

The development of information technology brings security database new challenges. Privacy principles are even being mandated internationally through legislations and guidelines, and this requires the security database to verify that it adheres to its declared data disclosure policy. An auditing system satisfies the above desiderata well, but only auditing the result of query is not enough, because malicious user can access sensitive information by inferring the result of query. This requires the auditing system to have the basic inference capacity. In this paper, a security database inference audit framework is proposed, which have the ability of MVD inference, FFD inference, and FD inference. At the same time, the auditing system keeps the properties of fast, precise, and fine-grained.