MapCube : a Mobile Focus & Context Information Visualization Technique for Geographic Maps : Location Based Service Context Model and a Prototypical Implementation for Ridesharing Scenarios Evaluated Using an Empirical User Study
A problem inherent in many mobile applications are the form factor restrictions imposed by mobile devices, that directly translate into limited screen real estate available to the application. In particular spatial tasks can be more difficult to perform. Examples for such tasks involving geographic map renderings are route planning or displaying Points of Interest (POIs), tasks that are often part of Location Based Service (LBS) applications, e.g. geo-fencing and car-navigation-systems. The direct consequence of limited screen real estate is that relevant information or POIs to be displayed by a navigation-system might be situated at off-screen locations, while the user is following navigation instructions, i.e. performing the Focus Task (FT) with the associated focus view occupying the whole screen and showing an appropriately zoomed map rendering. Geographic location context surrounding the viewed map portion is not visible. This thesis addresses this information visualization problem in light of the use case real-time ridesharing to allow the driver to be involved in the decision making process to identify passengers that are possible ridesharing-matches, and use previously neglected matching criteria, that require spatial tasks, to be considered. Based on the developed MapCube Focus and Context Model, that takes the LBS ridesharing into account, this thesis derives a conceptual model for a novel information visualization technique, the , as a solution. This solution addresses the challenge to reuse the screen for context while the focus view is displayed fullscreen, to maintain the focus view and associated mental model of the user, and at the same time display geographic context objects. These geographic context objects can be locations, polylines, arbitrary shapes and text, as well as spatial relationships. These objects can be visually linked to objects within the focus view, e.g. by drawing a line along the connecting route within the geographic map rendering. As an extension of folding of the spatial substrate in existing information visualization techniques, the proposed solution implements bending as an improvement of techniques for information spaces of large two dimensional extent, to allow the described screen reuse. The implemented prototype relies on 3D perspective projection and transparency as the basis for screen reuse. It demonstrates the use of depth-cues, which invoke preattentive visual processing to avoid composite-fusing of the transparent foreground layer with the 3D background, to improve perceptibility. The particular use of perspective projection and transparency allows for seamless transition between the focus view optimized for car-navigation, and the view, additionally revealing geographic location context. This is important in the ridesharing use case, for reacting to events from the ridesharing LBS, e.g. when passengers become relevant to a driver due to their proximity, making them candidates for a ridesharing-match. The prototypical implementation has been shown effective in a user study, to visualize as described above. In a comparison with a technique to display off-screen objects, subjects needed on average only half the time for counting off-screen objects around off-screen locations, and made only half the errors. This difference in population means is due to the varied independent variable visualization technique. This has been shown by a one-way analysis of variance (ANOVA). A statistically significant effect of the visualization technique can be reported with 95% confidence.
Nutzung und Vervielfältigung:
Alle Rechte vorbehalten