CALL FOR PARTICIPATION
AAAI 2000 Spring Symposium: Adaptive User Interfaces

As computers become more accessible, the problem of designing
effective user interfaces becomes more severe. Many new users expect
interacting with a computer to be as natural and intuitive as
interacting with a person, but current interfaces are artificial and
constraining.  One particular weakness with most interfaces is their
static nature. Programmers create these interfaces to interact
identically with all users and for a range of tasks, without
considering differences in knowledge, preferences, and purpose. At
best, some interfaces allow limited customization by explicitly
setting preferences and options.

A new cross-disciplinary approach has emerged in recent years from
researchers in human-computer interaction, information extraction,
machine learning, and other fields. In this approach, a system takes
advantage of feedback from its user or the environment to adapt its
performance. This adaptation can take a number of forms. Some examples
include: presenting different information to the user, altering the
presentation order, changing the level of interaction, or describing
information differently. Feedback is generally unobtrusive, with the
goal of making the interface more effective without the burden of
explicitly configuring the system. A popular approach is to model
facets of the system's functionality with parameters, such as a user,
task, or world model. An adaptive component estimates parameter values
from feedback, and a performance component uses these values to
control the information presented.

This symposium will invite researchers from relevant fields to meet
and discuss common issues and problems. We are particularly interested
in complete working systems, and encourage participants to demonstrate
their systems.

Submissions

Potential participants should submit a concise 1-2 page abstract of
their work in the area. The abstract should briefly describe their
specific problem, their formulation of the problem into a well-defined
computable task (including a description of models used, if any),
their method of obtaining feedback to adapt the system, how different
kinds of feedback affect system performance, and experimental results
or project status. Electronic submission is preferred. Email
submissions to Seth Rogers <rogers@rtna.daimlerchrysler.com> in
PostScript or PDF format. For more information, see the symposium web
page at www.isle.org/~aui.

Organizing Committee 

Seth Rogers (cochair), DaimlerChrysler Research and Technology Center
North America; Wayne Iba (cochair), Institute for the Study of
Learning and Expertise; Mathias Bauer, German Research Center for
Artificial Intelligence; Pat Langley, Stanford University (CSLI);
Christopher A. Miller, Honeywell Technology Center

Submissions are due October 8, 1999.