In this work, we propose to advance in the field of information security in the areas of ranking systems (RS) and control systems, using ideas from the area of information theory.
First, we propose a RS that groups users based on their preferences, introducing similarity measures. The system presents possibly distinct rankings for the same product in different user groups. Besides presenting more personalized rankings to users, it is a system more resistant to attacks than the state-of-the-art. We then explore the effect of bribing in reputation-based RS, in the usual scenario (a ranking for each product) and in the scenario we proposed. We find the optimal bribing strategies, and we evaluate our methodology with real data, being the devised ranking system more robust to bribery. Finally, in control systems, we present methods to find the placement of the minimum number of inputs in LTI systems and switched LTI systems, in the eventual scenario where a set of controllers may fail, e.g., due to a cyberattack. In the first case, we prove that the problem is NP-complete. We design algorithms to solve the problems explicitly, and also to approximate the solution in polynomial time.
Keywords: Ranking Systems, Control Systems, Information Security, Bribing.
Dr. Guilherme RAMOS has a Ph.D. (2018) in Information Security at Instituto Superior Técnico, Lisbon, Portugal, through the doctoral programme in Physics and Mathematics of Information. He received the B.Sc. (2011) and M.Sc. (2013) degrees in Applied Mathematics from the Instituto Superior Técnico, Lisbon, Portugal. His areas of interest are: Structural Systems; Control Theory; Cryptography; Information Theory; Ranking Systems; Recommender Systems; Consensus Algorithms.