Brief
Outline of GeoVISTA Studio
Mark
Gahegan & Masa Takatsuka
One
barrier to the uptake of Geovisualization and Geocomputation is that,
unlike GIS, there is no system or toolbox that provides easy access
to useful functionality. The experimental environment, GeoVISTA Studio,
attempts to address this shortcoming. Studio is a Java-based, visual
programming environment that allows for the rapid, programming free
development of complex data visualization, exploration and knowledge
construction applications to support geographic analysis. It achieves
this by leveraging advances in geocomputation, software engineering,
visualisation and machine learning.
Currently,
Studio contains full 3D rendering capability and has the following
additional functionality: interactive parallel coordinate plots, visual
classifier, sophisticated colour selection (including Munsell colour-space),
spreadsheet, statistics package, self-organising map (SOM) and learning
vector quantisation. By combining visual and computational approaches
within the same environment, many new forms of analysis become possible,
capitalising on both the pattern matching and computing abilities
of machines and the cognitive abilities of humans.
At
its heart Studio has a component-oriented software building system
(called “builder”) that employs a visual programming environment to
connect program components together into useful applications. The
builder allows different components, each offering pieces of the required
functionality, to communicate freely with each other. However, the
nature of these connections, i.e. what should be connected and how,
is not clear at the outset. Consequently, the system needs also to
provide an experimental environment to test and discover how components
should be connected to maximise the effectiveness of constructing
knowledge or otherwise analysing geographical data. Java Virtual Machine
Java APIs JavaBean capable Builder Bean A Bean B Bean C Bean D JavaBean
APIs A builder constructs an application by connecting program components.
In
order to carry out the sophisticated data analysis tasks outlined
above, a system has to bring together the various kinds of computational
tools and techniques in a co-ordinated fashion, with a large degree
of interaction. For example, clicking on a data string in the Parallel
Coordinate Plot will select the appropriate row in the spreadsheet,
and vice versa. In total, six different types of co-ordination are
offered between visualization and analysis components: (1) Metadata:
components share the same data descriptions. (2) Data: components
share the same dataset. (3) Sample: components share a sampling strategy.
(4) Selection: components share the same data selection.
(5) Focus: components share the same highlighted values. (6) Visual
Assignments: components share visual appearance.
Our
experiences so far in developing Studio applications indicate a promising
gain in efficiency over traditional programming methods, and a much
greater degree of integration and co-ordination among the component
pieces, fostering easier exploration and better understanding of both
tools and data. With the visual programming environment now completed,
future Studio development effort will focus on further tools for geographic
visualisation and analysis. Our current plans include interactive
scatterplots, Bayesian knowledge discovery agents, and metadata (including
semantic histories) to allow us to study and communicate the formation
of geographic objects in greater detail. Further information about
Studio, including sample images and downloadable applications and
data are available from http://www.geovista.psu.edu/studio/.
Future developments will also be posted to this site.