Comparing inference control mechanisms for statistical databases with emphasis on randomizing