HCI Project Ideas Discussion

1
HCI Project Ideas Discussion
By Bill Long
September 21, 2001
2
Discussion Overview
I will discuss 4 broadly defined ideas that fall
within the Modeling and Appropriate Action
components (of any situationaly aware system)

1. Patterns and Routines
2. The Perception of Annoyance
3. Progressive Alerts and Display Info
4. The Big Brother Effect?

01
3
Patterns and Routines
The strong impact of social environments in
shaping behavior and expectations allow us to
identify categories of situations, as well as
features and routines, from which we can begin
model peoples tasks pg. 4

• Focus - mobile users (phone, Palm)
• Goals - three (Modeling, Display, Sensing)
• Kinds - general situations and specific audiences
• Early formative evaluation stage of the study
  (pg. 11)

02
4
General Situations
Can we accurately predict appropriate
interruption types based on social environment
measures instead of task-specific or
attention-specific measures?
Situation Your cell phone resting on a table
top, turned on, and is visible to
others. Goals Modeling -Find the of time it
would be appropriate to ring compared to vibrate
or automatic voicemail. Sensing - Categorize
the situation (from a sensing perspective) Result
s Guidelines for appropriate interruption state
/ setting under certain patterns of conditions.
03
5
Specific Audiences
Can we accurately predict attention and task
state by mapping job-defined routines?
Situation A doctor, with a Palm, is entering new
data into an application for patient
records. Goals Modeling The task and
identifying / understand the situation Display
What information would be most valuable to the
user under certain task stages (what should I
show) Results A weighted list of for
interruption information at identified task
states Rules for predicting attention (based on
job-defined routines / interactions with the
device)
04
6
The Perception of Annoyance
How users / people come to perceive assistant
agents (interruptions) as annoyances.
Focus Microsoft Paperclip A focused study of
why Clippy failed Goal Identify the factors /
norms that were broken and how they led to the
the perception of annoyance. (to form
guidelines) Sub-Goals How do the factors relate
to one another The weight of each factor in the
perception of annoyance What To Study Timing
The timing of the interruption within the
framework of the task. Type in Info How
relevant is the information? How does that
affect perception / trust? How Interrupted
Full Attention vs. Awareness, Different
Graphical representations
05
7
The Perception of Annoyance
How users / people come to perceive assistant
agents (interruptions) as annoyances.

• Results
• An understanding of the Microsoft Paperclip
  failure (History Palm from Newton)
• Guidelines / Rules for how not to implement a
  interruption assistant.
• Timing Modeling
• Type of Info Display
• How Interrupted Display Interaction

06
8
Progressive Alerts and Information Display
How to effectively summarize information at
different display stages.
Situation It uses noninvasive displays to
summarize information about new messages, unread
mail and upcoming meetings, taking care that the
displays change slowly and do not otherwise
attract attention. pg. 1 Design Foci Type of
Alert (icon, kinetic type, text, sound,
color) Type Stage which type is appropriate
at each level ( is this determined by the
information needed) Information Display How do
different summaries affect the meaning Possible
Example How to effectively summarize email in a
progressive alerts display
07
9
Big Brother Effect?
Can a system be too visually predictive for
users? And how does the perception of the
interruption / device change as the visual
non-visual proportion changes?
Visual Non-Visual Visual these are
in-your-face interruptions, summaries, etc. that
require some level of attention /
interaction Non-Visual these are the
behind-the-scenes adjustments based on the
identified state of the user Foci Will the
users perception of the system change as the
system nears its ideal state (in the future)?
Can a system be too perfect of a predictor for a
user? Can a system be too visual? What is the
visual threshold and which types of visual
interruptions can be moved to non- visual? How
does the visual non-visual proportion affect
the perception of the system / interruption? Can
this be linked to metaphors for different
proportion states?
08