A hybrid-based filtering approach for user authentication

Abstract

Big data based user authentication is a new approach that leverages the power of Big Data analytics to develop a fertile field for the next generation user authentication. This new approach relies on “something you do”-based verification methods, where the users' dynamic behaviors are analyzed in order to generate real-time uniquely identifiable information about them. Once the unique user's identification is generated “authentication on demand” can be achieved through user challenging questions. In this paper, a new model is proposed to generate these users' identifiable information, where the main concepts of creating users' profiles in Recommendation Systems (RS) is used. RS are using the users' profiles to determine the user's preferences so that they can suggest a list of recommendations to other users with similar preferences. However, in the proposed model these profiles are employed to determine the user's personality traits that have a substantial influence on his/her identity verification. Based on these users' profiles the challenging questions will be issued only once to protect the users' responses from being compromised, and will be generated for the most recent user actions to help the legitimate user to easily remember and successfully complete the challenge.