Abstract: Due to derivations in current role definition from the changes of the bossiness process and information infrastructure,organizations are vulnerable to internal threat.A role dynamic adjustment algorithm is proposed based on the defensive idea of changing the set of roles within the organization regularly and reasonably.The algorithm defines an objective function with adjusting parameters to balance the two elements,which are the user privilege actual use data and the system administrator expert knowledge.Based on heuristic search strategy and sub-set pairing technique,a group of candidate roles are obtained.From these roles,a set of adjusting roles which can achieve a predefined score are obtained,by using a certain heuristic function.Finally,in order to reduce role redundancy,the users of the organization are reassign the roles from the adjusting roles,so getting a new Role-Based Access Control(RBAC) configuration.By using the audit logs from a hospital management system,the performance of the RDA is analyzed.The experimental results show that the proposed algorithm can efficiently adjust the RBAC configuration for the special organization,so it can provide concrete base for resisting the insider threats.

Key words: Dynamic adjustment, Heuristic search strategy, Insider threats, One class support vector machine, Role based access control