Exact Boolean matrix factorization based on concept lattice factors and its application in RBAC

Boolean matrix factorization methods provide representations of an object-variable data matrix by a product of two different matrices, one describing a relationship between objects and factors, and the other describing a relationship between factors and the original variables. The objective of this paper is to use indiscernibility matrix finding the matrices based on formal concept factors, and design an RBAC(Role Based Access Control) using formal concept analysis. For this purpose, we derive a formal context from the access control matrix that represents role-based access permission. We demonstrate the proposed method on an electronic medical record system of a hospital and some synthetic data, real world datasets. The analysis indicates that the proposed method can make the number of role mining relatively small and follows the RBAC constraints: static separation of duties and role hierarchy.