User Group Profiling through Mobile Application Usage Behavior

Smartphones come pre-loaded with a relatively similar set of applications (abbr. apps) for all users of that particular smartphone brand. They treat users as if they are all part of one big group. This is a simplistic supposition that all smartphone users are alike and have similar usage characteristics. Furthermore, mobile user interfaces are static and do not adapt to the way apps are used. The device maker determines the app icon and app groups on the pre-existing widgets and they stay intact in their current state regardless of the user’s behavior. In this regard, based on app usage behavior, there is a clear need to first identify the groups of user behaviors and then adapt the interface for each user group. In this paper, we propose an ML-based approach to discover users’ group behaviors based on their mobile application usage, with the objective of delivering an appropriate interface adaptation for each user group. To this end, we tested our method on a real dataset that we collected from real users over a one-month period. We discuss in this paper the discovered clusters of users’ behavior and we outline the parties who might benefit from our findings.