This is Me: A Bayesian Approach to Weighting Digital Identity Sources

Online accounts and services have become a ubiquitous facet of modern existence. As society becomes increasingly dependent on digital services to conduct the fundamental aspects of daily life, online identities may possess the capacity to provide verification that an individual is a ‘real person’ with a ‘true’ identity. Traditionally verification of identity has been achieved by cursory examination of paper documentation and simple matching of the attributes contained therein. Improved technology and printing capacity has since reduced confidence in this type of identity documentation, highlighting the need for methods better acquainted with the digital age. A recognized challenge in utlising online accounts as proof of identification is the various levels of reliability associated with different types of accounts. This variation is based on the importance of account and the processes of verification required to initially create the account. This paper explores the application of Bayes theorem to determine the reliability of online accounts based on the account type and the attributes it contains. Bayes rule is applied subsets of attributes contained within various types of online accounts that are presented as ‘identity sources’. This seeks to demonstrate that less reliable sources such as social media accounts will produce lower trust scores than more reliable sources such as online banking. The results for which will be used to determine if digital identity sources can be relied upon in place of traditional paper identification documentation. This research and testing has been conducted as an element of a larger intelligent identity authentication system that seeks to create a solution that proves an identity is genuine via an individual's digital footprint.