Abstract
Authentication is a cornerstone in secure systems aiming to restrict access to legitimate claimants only. Authentication systems can be generally classified into knowledge-based (e.g., passwords), token-based (e.g., credit cards), or biometric-based (e.g., fingerprints). In this paper, we discuss the strengths and weaknesses of each class of authentication approaches with an emphasis on privacy related issues. We survey and present the related literature showing a gap on addressing users privacy concerns. We propose a Bayesian network approach for addressing and modelling privacy factors. We discuss the preliminary evaluation of the proposed approach. Recommendations for making privacy features more tangible and suggestions for future research directions are discussed.