Longitudinal Analysis of Low-Level Web Interaction through Micro Behaviours

Abstract

Figure 1: Evolution of the duration of users' episode over time in each phase. ABSTRACTTo truly understand how people learn to navigate and use a Web site or application, we need to collect real usage data over extended periods of time. Detailed Web interaction data gathered in the wild (from URLs visited, to keystrokes and mouse movements) has the potential to provide an indepth, ecologically valid view of interaction, and enable an understanding of how behaviour evolves over time. Interpreting such data is extremely challenging, however. We present a longitudinal data-driven analysis of fine-grained interaction data captured from 14,000 recurrent users over 12 months. At the core of our approach is the aggregation of low-level interaction data into micro behaviours. By analysing changes in these behaviours as a function of users' accumulated interaction time, we were able to demonstrate how users' interaction evolves as they become more familiar with a Web page. The results demonstrate that monitoring micro behaviours offers a simple and easily extensible post hoc means of understanding how Web-based behaviour evolves over time.

Automatic detection of usability smells in web applications

“…Regarding the study of interaction logs, the work of Apaolaza et al (Apaolaza et al 2015) analyze low level logs captured "in the wild", i.e. from real users in real websites performing free tasks.…”

LogUI: Contemporary Logging Infrastructure for Web-Based Experiments

“…Examples of solutions from this period included MLogger [10], PooDLE [5], Search-Logger [18], Wrapper [13], UsaProxy [3,4], the framework by Hall and Toms [11], WHOSE [12], and YAS-FIIRE [20]. Some of these solutions required additional software to be installed (such as browser toolbars), while others made use of an intermediatary proxy server to inject logging code, as used in subsequent studies [2,6,7,15]. Despite advancements, these approaches were less than ideal [8].…”

Self-Refactoring: mejoras automáticas de usabilidad para aplicaciones web

“…En trabajos más recientes sobre el estudio de logs de interacción, la propuesta de Apaolaza et al (Apaolaza, Harper and Jay, 2015) analiza registros de eventos de bajo nivel "en el llano", es decir tomándolos de usuarios reales realizando tareas libremente en sitios reales. Ellos transforman estos logs en entidades de más alto nivel llamadas microcomportamientos, y las utilizan para demostrar cómo el comportamiento de los usuarios evoluciona con el tiempo.…”