Section: New Results

Research Methods

Participants : Caroline Appert, Michel Beaudouin-Lafon, Anastasia Bezerianos, Olivier Chapuis, Jérémie Garcia, Stéphane Huot, Ilaria Liccardi, Wendy Mackay [correspondant] , Emmanuel Pietriga.

Human-Computer Interaction is a multi-disciplinary field, with elements of computer science, software engineering, experimental psychology and anthropology. More recently, designers have joined the CHI community, offering an important perspective, but also a different fundamental research paradigm, which differs from the value systems of engineering and the natural sciences. We explored the paradigm of Research through Design [34] , which we differentiate from traditional epistemologies in the human sciences. We distinguish design from research-through-design: the end goal is not to produce an artifact, but rather to frame an alternative future and uncover unmet human needs, desires, emotions, and aspirations. We identified three research perspectives that have been adopted within the HCI community: Projection explores possible future states, Place specifies the context in which design artifacts presented to gather data, and Point-of-View identifies the philosophical perspective imposed by researchers. Our goal is to understand what it means to conduct research through design and how to value research-through-design contributions.

In addition to exploring general questions about research paradigms, we also explore more focused questions that apply new research methods to the design of multi-surface environments. Large interactive surfaces (like the WILD platform) are interesting collaborative settings that allow viewing of large amounts of visual information. Before we are ready to use these platforms in real visual analysis situations, where people can place themselves at different positions around the display, we must first understand how the perception of visual information is affected by perspective distortion introduced by varying viewing distances and angles. A deeper understanding of such distortion effects can help visualization researchers design effective visualizations for these spaces and implement interaction techniques to aid in extreme distortion situations. We conducted [16] two studies on the perception of visual variables that encode information, such as Angle, Area, and Length, and found that perception is impacted differently at different locations on the screen, depending on the vertical and horizontal positioning of this information. The visual variable of Length was the most accurately perceived across the display. Our second study examined the effect of perception when participants can move freely in such situations, compared to when they have a static viewpoint, and found that a far but static viewpoint was as accurate but less time consuming than one that included free motion. But we observed that in free motion participants often chose non-optimal walking strategies that can increased perception errors, thus we provide precise recommendations on where and how to move in such environments. This work is a first step towards understanding and predicting the impact of large display environments to people's understanding and tasks.

Annotations play an important role in visual analysis and record-keeping. We discuss the use of annotations on visualization dashboards citeelias:hal-00719221, collections of linked visualizations, focusing on business intelligence analysis through a user centered design process from expert analysts in this domain. The first contribution bridges the gap between expert analyst needs and designers, when it comes to visualization annotations. The second offers a novel approach to annotating of visualizations, “context aware annotations”: We annotate data queries directly, rather than image/chart locations. Annotations are present irrespective of the visual data representations users select (different charts, numeric tabular views of their data, etc). We focus particularly on novel annotation aspects made possible by this approach, such as multi-visualization annotations, annotation transparency across charts and data hierarchy levels, and the possibility of recommending to users annotations done to similar data to enable annotation re-use. We also consider new challenges that arise from such our approach, such as what happens to annotations when the underlying data is changed, and provide recommendations and design solutions.