View article

Empirical studies of user activity, based on data collected from the systems with which users interact, present technical challenges related to the transformation of data streams to a form suitable for analysis. In this paper we discuss the particular challenges confronted during a study of user interruption behaviour based on low-level “keystroke” data and the ways in which these challenges were addressed. We also report on a method of data cleaning and analytical preparation that was developed and consider its effectiveness and potential applicability for similar studies.