An extended memoryless inference control model: partial-table level suppression

Memoryless inference controls are an important class of inference control methods for online statistical databases. Other inference controls are usually too complex to use in online systems. Cell level controls have been shown to provide a low level of indentification risk along with a relatively high level of release of nonsensitive statistics, but are also too complex. Table level controls have a reasonable level of complexity. A higher level of control is desirable, however to preclude the necessity of using other less accurate control methods in conjunction with the table level controls. The authors present a method which allows the release of some statistics at a level below the table level of inference control, thus providing the release of a greater number of statistics with a comparable level of identification risk. The method provides user look-up tables to calculate the risk for a particular query