Visualization
ConceptsMackinlay, J.
[MAC86]
APT is a visualization system which is based on the generate-and-test
paradigm. It works with 2D discrete data presentations in business-graphics formats
such as bar, scatter and line charts. The graphics primitives include areas (such as
circles and bars), lines, marks (such as + ) and visual attributes such as color, size and
orientation. The primitives are organized into primitive graphical languages. Each of
these primitive languages uses one or several graphics primitives. The primitive graphical
languages include: horizontal axis, vertical axis, line chart, bar chart, scatter plot,
color, shape, size, saturation, texture, orientation, tree and network. The grammar for
generating alternative representations is defined by three composition rules, which
combine the primitive graphical languages into complete graphs. The rules are double-axis
composition, single-axis composition and mark composition.
Double-axis composition can compose graphical sentences that have identical horizontal and
vertical axes, such as showing the sales data by year for two different divisions on the
same set of axes. Single-axis composition aligns two sentences that have identical
horizontal or vertical axis. For example two bar charts which are side by side, one
encoding gross national product (GNP) by country and the other encoding population by
country. Country is the common axis, while GNP and population are the different second
axis. Mark composition uses two different attributes of the same area or mark to encode
different variables, such as using the size of a circle to encode a state's population and
the color of the circle to encode the average age of the state's population. For testing
there are expressiveness and effectiveness criteria. A set of
facts is expressible in a language if it contains a sentence that encodes all the facts in
the set and not encodes additional incorrect data. Effectiveness criteria can be based on
a number of different factors. For example a presentation is effective when it can be
interpreted accurately or quickly, or when it can be rendered in a cost-effective manner.
For each generated presentation, the primitives are filtered, with their expressiveness
criteria used to generate a list of candidate designs. The effectiveness criteria are used
to order the candidate designs so that the most effective presentation will be the first
choice.
APT was developed on a Symbolic LISP Machine using MRS, a representation and logic programming system. APT is a functional prototype which consist of a logic program and a rendering system. The logic program is about 200 rules, and the rendering system is about 60 pages of LISP code.
Last modified on March 29, 1999, G. Scott Owen, owen@siggraph.org