Characterizing Smartphone Usage: Diversity and End User Context

Introduction

In the past few years the mobile device and mobile services markets have been affected by a strong emergence of smartphones. Smartphones are mobile phones that offer advanced computing abilities and connectivity options. Smartphones are programmable mobile devices, running complete operating systems in a manner similar to traditional computers. These features enable new kinds of mobile services that in turn shape the usage habits of smartphone users. As smartphones provide more and more applications for an increasingly wider range of usage situations, they become an increasingly integrated part of users’ everyday life. The programmability of smartphones gives a possibility to turn the devices into data collection platforms providing detailed information on users’ smartphone usage habits. The devices’ strong integration into users’ everyday lives gives a possibility to infer information also about the users’ other, everyday, habits. We have used handset-based measurements to collect rich smartphone usage data directly from handsets of the users. The data are used to detect place-related end user contexts, and to extract smartphone usage session and application session information.

End user context is seen as important information when trying to develop more personalized mobile services and applications. For example, Gartner (2011) predicts that by 2015 context, particularly centered on location, presence, and social interactions, will be “as influential to mobile consumer services and relationships as search engines are to the Web”. Detecting significant places meaningful to a user and end user contexts has gained more and more interest also in the academia. The potential value of context information lies in its possibilities to predict presumable differences in user behavior and usage habits related to the different contexts/situations of the user.

Smartphone usage sessions capture the user behavior and usage habits of a smartphone user on a high level. By dividing the usage sessions into application sessions we can study the structure of the usage sessions, as well as provide insight into the usage of individual applications. With the relatively comprehensive handset-based data which enable us to produce time-wise linked context and usage information we can examine the effects of context on smartphone usage.

The purpose of this article is to study smartphone usage habits of our handset-based data collection participants (i.e., panelists) by first describing the higher level smartphone usage through analysis of different smartphone usage session characteristics. After the higher level study we pursue deeper understanding of the usage habits by moving to application session level and the usage of individual applications. In both phases we consider the usage first in general and then in different end user contexts to acquire insight of the contextual patterns in smartphone usage.

Mobile end user context has been studied before with different methodologies, but inferring it from handset-based data is still a relatively new approach. The same applies to handset usage studies. This is due to the fact that handset-based data collection could not exist before the capabilities of handsets were at an adequate level (i.e., smartphones were in use). Thus the body of previous research around mobile end user context and overall smartphone usage is relatively slim and the external validity of the results is questionable due to relatively narrow datasets. This article contributes in providing new results for comparison by using own datasets and original methodology for end user context detection and smartphone usage session extraction.

The article is organized as follows. First, we go through some background and related literature. A central part of this is introducing the terms smartphone usage session, application session and end user context and explaining how they are defined and understood in this work. Second, the methodology and data used are described. Then, the results concerning smartphone usage sessions and application sessions in general, and in different end user contexts, are introduced and discussed. The final chapter concludes the paper.