Explorable Visual Analytics, Knowledge Discovery in Large and High-Dimensional Data

Abstract

Visual analytic tools are invaluable in the process of knowledge discovery. They let us explore datasets intuitively using our eyes. Yet their reliance on human cognitive abilities forces them to be highly interactive. The interactive nature of visual analytic systems is facing new challenges with the emergence of big data. Massive data sizes are pushing against the boundaries of current visualization capabilities. Also the emergence of complex datasets is asking for new ways of navigation in the high–dimensional space. EVA (Explorable Visual Analytics) is an in-progress work for developing a web–based tool for visual exploration of large and complex datasets. EVA tries to handle large data sizes through utilizing local GPU resources and a novel client/server architecture. It also provides an easy navigation mechanism for exploring high–dimensional data. This paper presents our experiments in knowledge discovery with EVA, using US Census employment dataset as our testbed. We hope our experiences result in designing guidelines and techniques for the future visual analytic tools of the big data era.