Jason Crampton (ISG), Eduard Eiben (Comp Sci), Gregory Gutin (Comp Sci), Daniel Karapetyan (U. of Nottingham) and Diptapriyo Majumdar (Indraprastha Institute of Information Technology Delhi (IIITD), India) have received the best paper award at the ACM Symposium on Access Control Models and Technologies (SACMAT) 2022.
The ACM SACMAT 2022 Best Paper Award
SACMAT 2022 is the 27th Symposium on Access Control Models and Technologies, which was held online 8-10 June 2022.
Title - Generalized Noise Role Mining Award (Best Paper)
Abstract - Role mining seeks to compute a set of roles R, a user-role authorization relation UA and a permission-role authorization relation PA, given a user-permission authorization relation UPA, and is therefore a core problem in the specification of role-based authorization policies. Role mining is known to be hard in general and exact solutions are often impossible to obtain, so there exists an extensive literature on variants of the role mining problem that seek to find approximate solutions and algorithms that use heuristics to find reasonable solutions efficiently. In this paper, we introduce the Generalized Noise Role Mining problem (GNRM) – a generalization of the Min Noise Role Mining problem – which we believe has considerable practical relevance. In particular, GNRM can produce “security-aware” or “availability-aware” solutions. Extending work of Fomin et al., we show that GNRM is fixed parameter tractable, with parameter r+k, where r is the number of roles in the solution and k is the number of discrepancies between UPA and the relation defined by the composition of UA and PA. We also introduce a further variant of GNRM in which the accuracy of the solution is defined by the number of users and permissions that are affected, rather than the number of individual discrepancies k. We show that this variant of GNRM is also fixed-parameter tractable. We then report the results of our experimental work using general-purpose solvers to solve instances of GNRM. Our key findings are that security-aware role mining seems to be easier than availability-aware role mining, based on reasonable assumptions about UPA, and security-aware Role Mining introduces a similar number of discrepancies to Min Noise Role Mining.
Authors - Jason Crampton, Eduard Eiben, Gregory Gutin, Daniel Karapetyan and Diptapriyo Majumdar
Conference - SACMAT 2022
