How to use interaction logs effectively for usability evaluation
2008, Gerken, Jens, Bak, Peter, Jetter, Hans-Christian, Klinkhammer, Daniel, Reiterer, Harald
In this paper we argue for a combined approach of logging analysis and interview techniques to evaluate the usability of web distributed applications. Logging analysis has proven as an efficient and effective method to investigate websites usability. However it seems insufficient for more complex applications. We outline a case study of a library search system, in which users' information seeking behavior is investigated with both the analysis of log files and qualitative interviews with real users. We argue that such an approach should provide a more accurate usability assessment of the system, especially when users' performance in a long-run is investigated.
Blockbuster : a visual explorer for motion picture data
2007, Rexhausen, Sebastian, Demarmels, Mischa, Jetter, Hans-Christian, Heilig, Mathias, Gerken, Jens, Reiterer, Harald
In this paper we introduce our visual explorer "Blockbuster" as a contribution to the InfoVis Contest 2007. The system's development followed a user-centered design process and a design rationale considering not only the pragmatic qualities of the system, but also hedonic qualities like aesthetics or "joy-of-use". Apart from briefly outlining the employed visualization techniques, we will focus on Blockbuster's interaction design, which is aimed at facilitating the selection, combination and mutual filtering of visualizations under a consistent interaction paradigm. Blockbuster thereby demonstrates the potential of information visualization for end-user-centered applications that blur the boundaries of information visualization, visual information seeking and browsing.
Dynamic text filtering for improving the usability of alphasliders on small screens
2007-07, Büring, Thorsten, Gerken, Jens, Reiterer, Harald
Previous research has shown that Alphasliders are an effective tool for searching an alphabetically sorted list when only limited screen space is available for the graphical user interface. To improve user satisfaction, we propose equipping the widget with a novel text filter to dynamically limit the slider range. In this way, users are supported in locating target items and in identifying records that are missing. The results of a comparative user evaluation run on a Personal Digital Assistant showed that 8 out of 12 participants preferred the filter widget to the classic interface. We further suggest an enhanced Alphaslider design to speed up user interaction.
Longitudinal evaluation methods in human-computer studies and visual analytics
2007, Gerken, Jens, Bak, Peter, Reiterer, Harald
In Human-Computer studies and visual analytics, the majority of the evaluation methods applied, collect data at a single point in time, in form of cross-sectional data. In several studies numerous visualization tools were evaluated in controlled experiments. Although the experiments discovered valuable findings, certain drawbacks of the research method were expressed. The time constraints of one-time experiments reduce the amount of training which can be given to the participants.
Furthermore, when the studies tried to measure the insight derived from the visualization tools the time constraints didn't allow observing how these insights develop over time or their interdependency. Further problems of cross-sectional studies are well known, like the selection of appropriate tasks, the mostly extrinsic motivation of the participants, the influence of a laboratory environment compared to a realistic work setting and whether a visualization tool does meet the work requirements in the long run. In this position paper we argue for applying longitudinal research methods in human-computer studies as an extension to cross-sectional studies and present a first approach towards a methodological research framework. We suggest a set of research questions and performance measures that would be benefical for extending cross-sectional studies with longitudinal ones. We also describe in two case studies, in which only cross-sectional research methods were used, how they can improved by longitudinal methods.