Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!decvax!decwrl!ucbvax!SRI-AI.ARPA!LANSKY
From: LANSKY@SRI-AI.ARPA.UUCP
Newsgroups: mod.ai
Subject: Seminar - Automatic Design of Graphical Presentations (SRI)
Message-ID: <8605141721.AA17509@ucbvax.Berkeley.EDU>
Date: Tue, 13-May-86 14:40:54 EDT
Article-I.D.: ucbvax.8605141721.AA17509
Posted: Tue May 13 14:40:54 1986
Date-Received: Fri, 16-May-86 01:38:31 EDT
Sender: jason@ucbvax.BERKELEY.EDU
Organization: The ARPA Internet
Lines: 47
Approved: ailist@sri-ai.arpa


	     AUTOMATIC DESIGN OF GRAPHICAL PRESENTATIONS

                 Jock D. Mackinlay (MACKINLAY@SUMEX-AIM)
	     Computer Science Department, Stanford University

                              PLANLUNCH
   	 	        11:00 AM, MONDAY, May 19
         SRI International, Building E, Room EJ228 (new conference room)

The goal of the research described in this talk is to develop an
application-independent presentation tool that automatically designs
graphical presentations (e.g. bar charts, scatter plots, and connected
graphs) for relational information.  There are two major criteria for
evaluating designs of graphical presentations: expressiveness and
effectiveness.  Expressiveness means that a design expresses the
intended information.  Effectiveness means that a design exploits the
capabilities of the output medium and the human visual system.  A
presentation tool is intended to be used to build user interfaces.
However, a presentation tool will not be useful unless it generates
expressive and effective designs for a wide range of information.

This talk describes a theory of graphical presentations that can be used
to systematically generate a wide range of designs.  Complex designs are
described as compositions of primitive designs.  This theory leads to
the following synthesis algorithm:
    o First, the information is divided into components, each
      of which satisfies the expressiveness criterion for a
      primitive graphical design.
    o Next, a conjectural theory of human perception is used
      to select the most effective primitive design for each
      component.  An effective design requires perceptual
      tasks of low difficulty.
    o Finally, composition operators are used to compose the
      individual designs into a unified presentation of all
      the information.  A composition operator composes two
      designs when the same information is expressed the same
      way in both designs (identical parts are merged).

The synthesis algorithm has been implemented in a prototype presentation
tool, called APT (A Presentation Tool).  Even though only a few primitive
designs are implemented, APT can generate a wide range of designs that
express information effectively.


VISITORS:  Please arrive 5 minutes early so that you can be escorted up
from the E-building receptionist's desk.  Thanks!