A Game Theoretic Recommendation System for Security Alert Dissemination

A game theoretic recommendation system is presented that aims to improve security alert dissemination and evaluation. The system increases the efficiency of its users (e.g. system administrators), who operate under time and resource limitations, by improving the information flow from security experts. Combining security alerts and local information the system recommends a resource allocation strategy to users. This strategy is obtained by solving a security game between potential attackers and defenders (users) where the game takes vulnerabilities and local preferences as input parameters. Zero sum and fuzzy security game formulations are studied and compared to each other numerically. A fully functioning proof-of-concept prototype illustrating the underlying concepts is implemented as a flexible and platform-independent Web application.