Senay, Hikmet, Ignatius, Eve

VISTA: Visualization tool assistant

This text is partially taken from [SEN94]

"Vista is a knowledge-based system that helps scientists design visualization techniques. It generates a technique for a given data set and lets users modify the design interactively."

VISTA's design is based on research in graphical perception. It extends the design methodology of APT (A Presentation Tool implemented by Mackinlay), which is just for 2D graphics, to three dimensions. VISTA's primary function is to automatically generate an effective visualization technique design for a given data set. VISTA also allows users to interactively modify the design.

The architecture of VISTA is splitted in three main units:

  1. The first phase is the data unit, in which the user can select a set of interactive data manipulation operators. There are some dynamically organized menus from which the user can select data sets and data variables. Furthermore he can obtain a data subset from the database by selecting several data variables from one or more data sets. The order of selection defines a default importance ordering, though the user can redefine it.
  2.  
  3. The second phase is the design unit, in which the system generates a composite visualization technique in three steps. Once VISTA generates the composite visualization technique, the user can modify it within the design unit. The three steps are:
    1. In the first design step, the data decomposition module decomposes the data subset into a set of simple data partitions, each of which can be visualized by a primitive visualization technique. The partitioning is based on functional decomposition in relational database theory.
    2.  
    3. In the second step, the visualization technique selection module finds a primitive visualization technique for each partition using the expressiveness and effectiveness rules. To optimize the selection process Senay and Ignatius organized VISTA's search space for primitive visualization technique for the most important data partition.
    4.  
    5. In the final step, the visualization technique composition module combines a pair of component visualization techniques by applying an appropriate composition rule.
    6.  

    Once all primitive visualization techniques corresponding to the data partitions are combined, the design unit presents a design description of the composite visualization technique to the user for possible modifications. When the user terminates the design process, the design unit produces a customized visualization technique that encodes the entire data subset selected at the data manipulation stage.

  4.  
  5. The third phase is the rendering unit, which gets the primary input from the design unit. The rendering unit creates an image according to the design description by using appropriate rendering algorithms. For this the unit uses composition rules to unify the primitives (marks) created by each rendering algorithm.At last the rendering unit displays the image resulting from this process on the user's terminal. Also the rendering unit provides interactive facilities for image manipulation, for example:

If you want to read more about marks and mark composition you may click here for further information.


Visualization Concepts

Last modified on March 29, 1999, G. Scott Owen, owen@siggraph.org