Work in Progress: K-Nearest Neighbors Techniques for ABAC Policies Clustering

In this paper, we present an approach based on the K-Nearest Neighbors algorithms for policies clustering that aims to reduce the ABAC policies dimensionality for high scale systems. Since ABAC considers a very large set of attributes for access decisions, it turns out that using such model for large scale systems might be very complicated. To date, researchers have proposed to use data mining techniques to discover roles for RBAC system construction. In this work in progress, we consider the usage of KNN-based techniques for the classification of ABAC policies based on similarity computations of rules in order to enhance the ABAC flexibility and to reduce the number of policy rules.