Using Computer Behavior Profiles to Differentiate between Users in a Digital Investigation

Most digital crimes involve finding evidence on the computer and then linking it to a suspect using login information, such as a username and a password. However, login information is often shared or compromised. In such a situation, there needs to be a way to identify the user without relying exclusively on login credentials. This paper introduces the concept that users may show behavioral traits which might provide more information about the user on the computer. This hypothesis was tested by conducting an experiment in which subjects were required to perform common tasks on a computer, over multiple sessions. The choices they made to complete each task was recorded. These were converted to a \u27behavior profile,\u27 corresponding to each login session. Cluster Analysis of all the profiles assigned identifiers to each profile such that 98% of profiles were attributed correctly. Also, similarity scores were generated for each session-pair to test whether the similarity analysis attributed profiles to the same user or to two different users. Using similarity scores, the user sessions were correctly attributed 93.2% of the time. Sessions were incorrectly attributed to the same user 3.1% of the time and incorrectly attributed to different users 3.7% of the time. At a confidence level of 95%, the average correct attributions for the population was calculated to be between 92.98% and 93.42%. This shows that users show uniqueness and consistency in the choices they make as they complete everyday tasks on a system, and this can be useful to differentiate between them. Keywords: computer behavior users, interaction, investigation, forensics, graphical inter-face, windows, digital Keywords: computer behavior users, interaction, investigation, forensics, graphical inter- face, windows, digita