Role Mining Heuristics for Permission-Role-Usage Cardinality Constraints

Role-based access control (RBAC) has become ade factostandard to control access to restrictedresources in complex systems and is widely deployed in many commercially available applications,including operating systems, databases and other softwares. The migration process towards RBAC,starting from the current access configuration, relies on the design of role mining techniques,whose aim is to define suitable roles that implement the given access policies. Some constraintscan be used to transform the roles automatically output by the mining procedures and effectivelycapture the organization’s status under analysis. Such constraints can limit the final configurationcharacteristics, such as the number of roles assigned to a user, or the number of permissions includedin a role, and produce a resulting role set that is effectively usable in real-world situations. In thispaper, we consider two constraints: the number of permissions a role can include and the numberof roles assigned to any user. In particular, we present two heuristics that produce roles compliantwith both constraints and evaluate their performances using both real-world and synthetic datasets.