Image Processing Reference
In-Depth Information
28.2.3 Software Scalability
A commonly held best practice in active data visualization begins with a visualization
that summarizes a large data set followed by a subsetting to examine its detail. This
practice requires active data visualization software that can execute visual queries
and scale to data sets of varying sizes. Software scalability includes the generation
of new algorithms that scale to the ever-increasing information sets that we generate
today. We wish to avoid the hidden costs that arise when we build and maintain
monolithic, non-interacting, non-scalable software models.
28.2.4 Information Fusion
Information fusion includes the capability to fuse the relevant information from
divergent multi-source multi-dimensional time-varying information streams. This is
the grand challenge problem in visualizations. Researchers must not just produce new
visual representations and data representations for specific data types or information
streams, but we must develop methods that fuse the relevant information into a single
information space and develop new visual metaphors that allow the analyst to
look
inside
this complex, multi-dimensional, time-varying space.
We must also develop techniques to
measure
scalability so new tools can be ana-
lyzed for their applicability in this domain. We must establish metrics that allow us
to evaluate both visual metaphors and data representations as they apply to scalable
algorithms. The best measurement will not only evaluate the representations accord-
ing to scale, but also to the number of insights, actions, or value achieved for the
analyst.
28.2.5 Technology Needs
What is needed in the visualization research agenda is to extend the state-of-the-art
visual and data representations to be able to explore the heterogeneous multi-source
multi-dimensional time-varying information streams. We must develop new visual
methods to explore massive data in a time critical matter. We must develop new tech-
niques for information fusion that can integrate the relevant pieces of information
from multi-source multi-dimensional information. We must develop new methods
to address the complexity of information, and create a seamless integration of com-
putational and visual techniques to create a proper environment for analysis. We
must augment our methods to consider visual limits, human perception limits, and
information content limits. Therefore, the following challenges can have significant
impact on science, engineering, discovery, and society:
Search WWH ::
Custom Search