This proposal describes an investigation of the fundamental display and interaction abilities of a personal computing device (PCD), a new class of non-traditional computers that have proliferated in recent years. A PCD is typically a low cost, portable, low power consumption device with unique display and user interface properties. Examples include personal digital assistants, next-generation cell phones, and portable game-playing devices. Although these platforms are built with general-purpose CPUs, color displays, user interactivity, and access to high capacity storage, many researchers question their usefulness outside their original application domain. The PIs will address this issue by investigating: (1) how to model a PCD's input properties to evaluate the effectiveness of user interactions, and (2) how a PCD's output properties affect a viewer's ability to perceive the information it displays. Results will be used to extend a PCD's limited input and output capabilities. These techniques are further analyzed in the context of a new application environment: the visualization and exploration of large, multidimensional datasets.
The PIs will address the limited input capabilities of a PCD with new interface modeling techniques. These techniques take as input a set of tasks and the constraints specific to the PCD, then generate a set of interaction operations that optimize performance for those tasks on that device. This work will investigate several open research problems: representations for relationships between low-level operators based on device dependencies; search techniques for constructing and evaluating operator combinations to match specific tasks; and interaction mechanisms that facilitate the execution of operators in the context of different tasks.
What a PCD can display depends on its resolution and screen size. These properties impose limits on the type and amount of information a given PCD can represent. This relationship will be studied with perceptual glyphs that try to maximize their information content by dynamically modifying their appearance. Experiments will be conducted to determine how the visual properties of a glyph can best be used to harness the abilities of the low-level human visual system. Limited screen space will be further extended by: (1) structuring a display to allow a smooth tradeoff between a glyph's size and its information content, and (2) assisted navigation techniques that help viewers locate, organize, and explore areas of interest in their data.
Experimental results will be used to build a working prototype of a multidimensional visualization system. This system will run on an existing PCD, and will be designed to include functionality that matches or exceeds what is available from similar programs running in a traditional computing environment. The prototype will be evaluated with usability experiments that ask domain experts to visualize, explore, and analyze data from their real-world research projects.
Mike Romeo (MS candidate)