Studies on Multi-objective Role Mining in ERP Systems

A common concept to ensure the security of IT systems, in which multiple users share access to common resources, is Role Based Access Control (RBAC). Permissions, which correspond to the authorization to perform an operation on a data or business object are grouped into roles. These roles are then assigned to users. The corresponding optimization problem, the so-called Role Mining Problem (RMP), aims at finding a role concept comprising a minimal set of such roles and was shown to be NP-complete. However, in real-world role mining scenarios, it is typically the case that, besides the number of roles, further key figures must be consulted in order to adequately evaluate role concepts. Therefore, in this paper, the RMP is extended to a multi-objective (MO) optimization problem. Potential optimization objectives are discussed in the context of Enterprise Resource Planning (ERP) systems. Furthermore, it is shown, how evolutionary algorithms for the RMP can be adapted to meet the requirements of MO role mining. Based on this, the integration of different optimization objectives is examined and evaluated in a series of experiments.