Approximate reduced model based on mutual information for security assessment

Choosing appropriate assessment elements is vital in security assessment, current criteria often result too many points, which influences the feasibility and correctness of assessment model. The general attributes reduction algorithms usually generate more than one results, in security assessment how to choose the result according these algorithms is not given. Based on the theory and methods of rough set, an attribute approximate reduction algorithm for information security assessment is proposed. The redundancy is measured with mutual information and redundancy synergy coefficient. Through calibrating the threshold parameter, different scales of reduced attribute set can be obtained according to specific application, which can make security assessment model more concise and effective.