Graph based strategies to role engineering

Abstract

Role Engineering for Role Based Access Control (RBAC) has emerged as a challenging area of research, both in industry and academia. The problem originates from the practical need to create a set of roles that accurately reflects the internal functionalities of an enterprise. Existing approaches that have used data mining techniques for this problem often generate too many candidate roles and do not consider the effect of a given combination of roles on the overall configuration.Identification of an ideal RBAC solution is only possible with a clear and concise evaluation of the RBAC configuration goals. To address this issue, we discuss use of the graph model for the Role Engineering problem and show how effective this approach is in the search for a role engineering solution. We evaluate and formalise the problem of identifying the minimum number of descriptive roles for RBAC using a graph model and propose how its variations can be represented. Finally, we introduce novel strategies using the proposed models for future innovation and perform experimentation on both real and synthetically generated data.