Controlling logical inference in multilevel database systems

• Authors: Matthew Morgenstern
• Venue: SP'88 Proceedings of the 1988 IEEE conference on Security and privacy
• Year: 1988

The potential for logical inference of high level information based upon lower level visible data presents an interesting and challenging threat to multilevel security. Such compromises of security are rather novel since they circumvent traditional security mechanisms and rely on a user's knowledge of the application, which is external to the security layers of the system. The potential for such inferences, and the multiple consequences of a corrective action, substantially complicate the task of classifying the data in a secure manner. Computer-based tools will be needed to assist in this process, especially when multilevel databases of substantial size and complexity are considered. Heretofore, the problems of inference and security have been amorphous and difficult to circumscribe. This paper proposes a framework for studying these inference control problems, describes a representation for relevant semantics of the application, develops criteria for safety and security of a system to prevent these problems, and describes the functionality of the proposed classification tool in terms of a scenario for its use.