Most Visual Data Mining (VDM) methods have been designed for PC hardware
using 2D graphics or 3D graphics on a monitor. VDM in immersive 3D Virtual
Reality (VR) systems using CAVE or Panorama arenas for visualization has so
far been dependent on special and expensive hardware, in order to achieve
real-time response to user interaction and navigation.
With the continuous improvements in computer technology, it is now possible
with standard high-end PCs to drive VR systems. They can visualize many
objects while simultaneously navigating among them and perform intensive
background calculations, all in real-time.
However, while the main principle behind the design of traditional VDM
methods is that e.g. 3D Scatter Plots are viewed from the ``outside-in'',
the immersed VR users can navigate around inside a 3D Virtual World (VW),
which may then also be viewed from the ``inside-out''. VR applications
provide more comprehensive input to the human senses and can therefore make
more efficient use of the human perceptual skills, when exploring large
datasets.
Our hypothesis is that a complementary and valuable benefit is achievable
concerning the detection of e.g. non-linear relationships in data, which are
most likely to escape traditional methods of data analysis.
To support our hypothesis, we have implemented visualization tools for
exploring data in VR, which are open-source.
|
