The inference problem

Abstract

Access control models protect sensitive data from unauthorized disclosure via direct accesses, however, they fail to prevent indirect accesses. Indirect data disclosure via inference channels occurs when sensitive information can be inferred from non-sensitive data and metadata. Inference channels are often low-bandwidth and complex; nevertheless, detection and removal of inference channels is necessary to guarantee data security. This paper presents a survey of the current and emerging research in data inference control and emphasizes the importance of targeting this so often overlooked problem during database security design.