CARS-AD: A Context-Aware Recommender System to Decide about Implicit or Explicit Authentication UbiHealth

Mobile devices using traditional authentication processes are vulnerable and inadequate. New approaches must be considered for solving this problem: environmental characteristics, device limitations and information obtained from sensors. This paper presents a recommendation system approach for information systems based on user behavior and information context in which the users are located. The recommendation system has been defined and deployed through filtering processes (content-based, collaborative and hybrid). The behavior is defined by the events and the actions that comprise the user activities. The experimental results indicate: (i) a more dynamic and autonomic mechanism for authenticating users in a pervasive mobile environment, and (ii) an efficiency improvement in detecting anomalies on authentication by using a similarity model and space-time permutation.