The use of an extended data model which represents both integrity and secrecy aspects of data is demonstrated. This Semantic Data Model for Security (SDMS) provides a technique that assists domain experts, security officers, and database designers in first understanding their security requirements, and then translating them into a good database design. Identifying security requirements at this semantic level provides the basis for analyzing the security requirements and the database design for inference and signaling vulnerabilities. Another contribution is a comprehensive taxonomy of security-relevant data semantics that must be captured and understood to implement a multilevel secure automated information system.>

Data security requirements go beyond traditional access control considerations to include relationships among data that must be protected. The author extends the data-modeling concepts to include the secrecy properties of data. The semantic data model for security (SDMS) is proposed as the vehicle for users, database designers and security officers to analyze and understand the security requirements of the application domain. The model aids in analyzing data security requirements and database designs by providing an easy-to-comprehend representation scheme. SDMS can represent a full range of security semantics of an application.>

A database system must have knowledge of the semantics (the properties) of the data it manages to accomplish its tasks. For a multilevel secure database system to provide effective multilevel support to users, it must have knowledge of the security-relevant data semantics. The use of an extended data model that represents both integrity and secrecy aspects of data is presented. The technique can be used as a database design tool and, more importantly, as a vehicle by which domain experts, database designers, and security officers can precisely define the security requirements for an application domain. A second contribution is a comprehensive taxonomy of security-relevant data semantics that must be captured and understood to implement a multilevel secure automated information system.>

The Second Rome Air Development Center (RADC) Database Security Workshop held in May 1989, featured a 'Homework Problem'. The Homework Problem was a 22 page specification for a multilevel secure (MLS) database. Solutions to the Homework Problem were provided by many of the major research and development efforts in MLS database systems. The author describes the requirements of the Homework Problem and summarizes the database design approaches taken to meet those requirements.>