Sociosensor: Measuring user behavior and experience with mobile devices

I. Mulder¹, G.H. ter Hofte¹ and J. Kort²

¹Telematica Instituut, Enschede, The Netherlands
²TNO Information and Communication Technology, Groningen, The Netherlands

Man is a social being, continuously and dynamically adapting to his social context. Current research into context-aware and mobile computing stresses the relevance of using context information in applications. However, despite occasional design successes, researchers are still lacking a systematic understanding which context information is relevant in what kind of situation and which kinds of services.

We exploit mobile devices such as PDAs and smartphones as an instrument for the scientific investigation into the dynamics of social phenomena. Mobile devices tend to travel along with people wherever they are and whatever they are doing, and consequently enter various social contexts. This literally puts these devices in an ideal position to capture several aspects of social phenomena.

In this paper, we describe Sociosensor, an extensible toolkit that exploits the hardware sensors and software capabilities of contemporary mobile devices to capture objective data about dynamic social interaction and social context together with subjective data about social experiences. We focus on software-based sensors that capture objective data about social context factors, such as proximity, communication and relation, using technologies currently widely available to programmers of smart mobile client applications, such as infrared, Bluetooth, WLAN, audio microphone, GPS, GSM cell-ID, phone/E-mail/IM logs and contact lists. In addition to software that logs objective information, we focus on experience sampling software that can query the user at multiple pseudo-random times for subjective information such as opinions and feelings.

Thus, we provide social sciences with an instrument to gain a much deeper, detailed and dynamic insight into these phenomena and their relations, which in turn informs the design of successful context-aware applications. In addition to such formative evaluation for design, the Sociosensor toolkit can also be extended with modules allowing summative evaluation of application usage in context. Sociosensor not only strengthens this crucial reciprocal link between evaluation and design, but also provides benefits to a broader audience of scientic communities including medical/biological sciences such as epidemiology.

Paper presented at Measuring Behavior 2005 , 5^th International Conference on Methods and Techniques in Behavioral Research, 30 August - 2 September 2005, Wageningen, The Netherlands.

© 2005 Noldus Information Technology bv