Intelligent User Interfaces
Annotated Bibliography
Index of Articles in this Bibliography
- Intelligent User Interfaces, by William E. Hefley and Dianne Murray
- SAGE, various authors
- Interactive Sketching for the Early Stages of User Interface Design, by James A. Landay and Brad A. Myers
- A Problem-oriented Classification of Visualization Techniques, by Stephen Wehrend and Clayton Lewis
- Intelligent User Interfaces: Issues, Approaches, Evaluation, by Loren Terveen, AT&T Bell Laboratories
- Intelligent Multi-Media Interface Technology, by Jeannette G. Neal and Stuart C. Shapiro
- User and Discourse Models for Multimodal Communication, by Wolfgang Walhster
- The Contributing Influence of Speech and Interaction on Human Discourse Patterns, by Sharon L. Oviatt and Philip R. Cohen
- An Intelligent Interface Architecture for Adaptive Interaction, by Sherman W. Tyler, et al.
- General User Modeling: A Facility to Support Intelligent Interaction, Robert Kass and Tim Finin
- Intelligent Interfaces as Agents, by David N. Chin
- Structuring Programs to Support Intelligent Interfaces, by Pedro Szekely
Intelligent User Interfaces
William E. Hefley and Dianne Murray
Proceedings of 1993 International Workshop on Intelligent User Interfaces
In this paper, the authors take different research ideas and place them into
a structure. They address design and implementation issues.
Much of the work which has been done in the field of intelligent interfacing
has been done in dialogue understanding and user modelling to support user
assistance, thereby providing usable presentations, error remediation, and
tutoring. There has also been work done on adaptable interfaces.
There are two issues which must be considered in order to make these types
of systems work effectively: How can we model the user/operator in complex
time-critical environments; and how can this knowledge be worked into the
design of interfaces to enhance human-computer interaction and overall
system effectiveness? There are difficulties inherent in dynamic modelling:
how to get information, how the information should be represented, and how
effective dialogue can be devised using the information. There are also
privacy and ethical issues which this paper did not discuss.
Many complex systems demand that the operator work in two capacities: as a
supervisor when the system is operating normally, and as a manual controller
during emergency or degraded conditions. Both of these require the operator
to have knowledge of the integration of the system as a whole as well as the
details of its functions.
There are five levels of interaction in human-computer systems:
- Functional Purpose. System and user goals, intentions.
- Abstract Function. Task structure; work organization.
- Generalized Function. Applications and functions.
- Physical Functions. Screen displays; I/O devices.
- Physical Form. Ergonomic and environmental.
To reduce the complexity of a user's role in complex tasks, either the user's
skill should be increased or the task should be made more simple. It is
very important to prevent errors in high-performance systems. Moreover,
providing a way to improve the user's skill is important in providing users
with guidance so they can think clearly and knowledgably about a problem and
work effectively with that understanding.
Different groups of people interact differently with computer systems.
Often, users are classified as experts or novices, frequent or infrequent
users. This is a consideration which must be thought about when designing
effective interfaces.
There are three potential roles of the user interface:
- To assist in the correct and effective useof the system's capabilities,
- To be proactive in the user's problem solving process,
- To provide training.
This paper gounds its design and implementation in user interface management
systems (UIMS). UIMS are high-level interactive software applications
intended to design, specify, prototype, execute, evaluate, and maintain
end-user interfaces. The UIMS has three components between the user and the
application program:
- Presentation. To provide the presentation of the interface to the user.
- Dialogue control. To control the structure or sequencing of the dialogue between the user and the application program.
- Application interface model. To define the semantics of the functionality of the underlying application.
SAGE
- SageTools: An Intelligent Environment for Sketching, Browsing, and Customizing Data-Graphics
Steven F. Roth, et al.
CHI '95 Mosaic of Creativity
- Interactive Graphic Design Using Automatic Presentation Knowledge
Steven F. Roth, et al.
Human Factors in Computing Systems, CHI'94-4/94
- SageBook: Searching Data-Graphics by Content
Mei C. Chuah, et al.
CHI '95 Mosaic of Creativity
- Data Characterization for Intelligent Graphics Presentation
Steven F. Roth and Joe Mattis
CHI '90 Proceedings
Synposis
Data-graphics, which provide a lot of valuble information in a visual format,
can be quite difficult to create. There are software packages available
which create data-graphics, but these packages have several faults,
including the availablility of too many options, ineffective "non-integrated"
displays, and difficulty in use for a person who is not an experienced
graphic designer.
The SAGE program itself is complex, but two interactive design tools,
SageBrush and SageBook, shield the users from the complexity of SAGE.
SageBrush allows the user to sketch designs using primitives. These
sketches are used by SAGE to determine roughly what the final data-graphic
should look like.
SageBook is an image-retrieval
system for use with the data-graphics creator, SAGE. The query interface in
SageBook does not require that the user learn specific data-graphic terms.
The user instead enters information in terms of spaces, disciplines,
graphemes (graphic elements), and grapheme properties. There are also
different layouts of graphs, such as charts, tables, maps, networks,
indentations, and grids. Information entered by the user is passed to a
data-matcher and a graphic-matcher (these matching functions are not
significant in terms of intelligent interfaces).
Significance
The SAGE system is important for intelligent interfaces because it is a system
which designs intelligently created displays. The actual interface in SAGE
did not seem to be intelligent, but the displays created were. When the
interface does act intelligently (by telling the user how something might
look better, or presenting options which have been specifically picked out
for a particular operation), it is acting on the information determined by
the program.
Intelligent Interface
In SAGE, the intelligent display of data-graphics is the intelligent interface
(SAGE's actual interface is not intelligent).
Interactive Sketching for the Early Stages of User Interface Design
James A. Landay and Brad A. Myers
CHI '95 Mosaic of Creativity
Synopsis
This paper discussed a system, SILK (Sketching Interfaces Like Krazy), which
allows designers to quickly draw out an interface. This allows prototypical
interfaces to be created rapidly by using common gestures in sketch drawing
and editing. The system takes the designer's sketch and transforms it into
an interface.
Significance
The system allows the user to create an interface, but the interface created
is not intelligent.
Intelligent Interface
The intelligence here is in the multimodal communication between the user and
the system. The user is able to draw sketches and use gestures, which the
program understands and interprets into a working interface.
A Problem-oriented Classification of Visualization Techniques
Stephen Wehrend and Clayton Lewis
IEEE 1990
Synopsis
This article discusses scientific visualization, in which users of a system
may visualize their application domain. Progress would be accelerated if
workers could use these techniques in their work. There are many ways to
visualize data with different qualities. For example, in a system where
there are objects which are round or square, white or red, and short or tall,
the worker may wish to compare the red items to the white ones, or the short
ones to the tall ones. One way to determine the appropriate method of
display is to anayze the user's goals.
Significance
This was an interesting article. Intelligent interfaces and visualization
techniques could definitely impact how workers do their tasks.
Intelligent Interface
Intelligent visualization of data is what gives this system its intelligent
interface.
Intelligent User Interfaces: Issues, Approaches, Evaluation
Loren Terveen, AT&T Bell Laboratories
(Slides)
Synposis, Significance, and Intelligence
These slides presented an excellent working knowledge as to what intelligent
interfaces are. By their nature, slides are notes, so it would be redundant
to describe what the slides discussed.
Articles from Intelligent User Interfaces
Joseph W. Sullivan, and Sherman W. Tyler, eds.
(C) 1991 ACM Press
Intelligent Multi-Media Interface Technology
Jeannette G. Neal and Stuart C. Shapiro
Synopsis
This project attempts to apply artificial intelligence to HCI technology which
would incorporate speech input, speech output, natural language text, graphics,
and pointing gestures. The objective is to simplify communication with
complex systems by making the interaction more natural. The authors create a
proof-of-concept prototype called CUBRICON. With this, you could, for example,
say "Change the color of this to blue" while pointing to 'this'. The
combination of visual, tactile, visual, and gestural communications is
referred to as the unified view of language. The system produces
relevent output using multimedia techniques.
CUBRICON is a system for Air Force Command and Control.
Several knowledge sources are needed for CUBRICON to perform its tasks:
- A lexicon - collection of morphemes, tokens, and signals that carry meaning in a given language
- A grammar defining the language used by the system for multimedia i/o
- A discourse model - key factors are continuity and relevance
- A user model - including level of expertise, perspective based on user's
role, user's value system, degree and nature of impairdness due to fatigue
or illness, and preferences concerning mode of communication.
- A knowledge base of output planning strategies to govern the composition
of multimedia responses to the user
- A knowledge base of information about generally shared world knowledge
- A knowledge base of information about the specific task domain of tactical
air control
Each of these sources are explained in detail in the article.
The actual working of the program is quite interesting. The user can, for
example, say, "Is this a steel plant?", and CURBICON might use speech to say,
"No, it is a munitions factory." Or, the user can ask, "Where is the Dresden
airbase?", and the computer would respond (with speech), "The map on the color
graphics screen is being expanded to include the Dresden airbase." The computer
would then say, "The Dresden airbase is located here," as the Dresden airbase
icon and a pointing text box blink.
Significance
This is an excellent example of an intelligent interface. Its user model is
especially well developed, showing maps with different icons to different types
of people (the icons shown are determined by the type of person who wants to
view them. For example, the system would not show country cottages to a
military mission planner).
Intelligent Interface
This system supports multimodal communication as well as intelligent data
display.
User and Discourse Models for Multimodal Communication
Wolfgang Walhster
Synopsis
The author starts by stating that in human-human conversation there is a frequent
use of gestures (especially pointing) to accentuate the meaning of what is being said. The point of
the paper is how multimodal (talking as well as gesturing) communication can
influence the user model and discourse model of an intelligent interface.
One of the problems with gesturing is that while it is natural, it could also
be ambiguous or vague. The article goes on to discuss various applications of
gesturing, as well as ways to implement its use.
Significance
The article was interesting, but it is not of much use to my research, since I
won't be working on a system which uses gestures (although that would be
very interesting).
Intelligent Interface
Multimodal communication; specifically, gesturing
The Contributing Influence of Speech and Interaction on Human Discourse Patterns
Sharon L. Oviatt and Philip R. Cohen
Synopsis
This paper studied various methods of multimodal communication:
- Speech Example: Audiotape monologue
- Interaction Example: Keyboard interaction
- Interactive Speech Example: Telephone dialogue
It presented the methods, results, and interpretations from the study.
Significance
The study was interesting to read, and is important to intelligent interfacing
(speech communication between the user and the system would make interfaces
easier to use and more friendly).
Intelligent Interface
Speech communication
An Intelligent Interface Architecture for Adaptive Interaction
Sherman W. Tyler, et al.
Synopsis
Discusses CHORIS, the Computer-Human Object-oriented Reasoning Interface System, a generic architecture
for intint developed at Lockheed. The current CHORIS domain is an emergency crisis
management system. This system uses intelligent display of data and accepts
input as natural language. It also has a Plan Manager, which assists the
user through its knowledge of typical plans for high-level goals.
Significance
This is a very interesting system to look at because it contains a lot of
ideas that I can use in my program.
Intelligent Interface
Intelligent data display, plan recognition, accepts natural language
General User Modeling: A Facility to Support Intelligent Interaction
Robert Kass and Tim Finin
Synopsis
This article discusses the notion of a "general user model" which could
be applied to a variety of systems. The authors present an extensive
diagram which shows many uses of the user model (unfortunately, the
diagram wouldn't scan well, so
here it is in list format). The authors also designed a system
called GUMS (General User Modeling System).
Significance
The user model is one of the most important components of an intelligent interface.
This article presented very good applications of user models.
Intelligent Interface
User modeling
Intelligent Interfaces as Agents
David N. Chin
Synopsis
This article discusses UCEgo, the intelligent agent component of UNIX Consultant,
a system used to help users with problems encountered while working with UNIX.
The system also recognizes user goals, which are divided into themes, plans,
subgoals, and metagoals; and user misconceptions.
Significance
This is a very interesting system. It takes into consideration things such as
acting ethically and being polite to the user. It is particularly interesting
to see the examples given when a user has a misconception, which the system
clarifies.
Intelligent Interface
Natural language, goals
Structuring Programs to Support Intelligent Interfaces
Pedro Szekely
Synopsis
This article discusses structuring programs so that they can effectively use
intelligent interfaces. The author gives an example of a chess program. The
user uses a mouse to select and drag pieces. The system is able to tell the
user what moves are legal or illegal.
Significance
The article didn't seem to have too much significance. It appeared as if the
chess example wasn't very intelligent, but maybe that's a case of things not
seeming intelligent once they're accomplished (see introduction).
Intelligent Interface
No actual IntInt techniques used, but examples given demonstrate how to work
IntInt techniques into existing programs.
Go back to the main IntInt page