A Data Science Approach Based on User Interactions to Generate Access Control Policies for Large Collections of Documents

Le résumé fourni par la source

A major concern of cloud storage users is the loss of control over the security and privacy of their information. The use of reliable and efficient access control schemes has proved to be mandatory for organizations whose operation involves sharing digital resources that must be kept private. However, defining access control policies (ACPs) is a non-trivial and critical process that is commonly delegated to expert humans. In this chapter, we propose a methodology based on data science composed of two main phases to comprehensively address the problem of generating APCs for attribute-based encryption schemes. Generated policies can be used to secure interactions with a large set of documents in cloud scenarios. The first phase is related to the detection of relevant users to spread information; the second phase focuses on the generation of policies based on propagation processes. In addition, we propose a strategy to analyze and evaluate the generated policies. The main contribution of this chapter is a method to spread ACPs through a graph of user interactions. We have tested our approach from real data obtained from the Instagram social network. The results obtained showed that the proposed strategies can effectively generate the ACPs to every user in an interaction graph from a small set of relevant users. Our proposal reduced in 96% the manual effort required to assign an ACP to each user in the tested dataset. Results also showed that generated ACPs are 92% effective when measuring the accuracy of the policies.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.